QASPER_compression_alignment_dev.jsonl

{"text": "Acknowledgements\tThis work was supported by Contracts HR0011-15-C-0113 and HR0011-18-2-0052 with the US Defense Advanced Research Projects Agency (DARPA). Approved for Public Release, Distribution Unlimited. The views expressed are those of the authors and do not reflect the official policy or position of the Department of Defense or the U.S. Government."}
{"text": "Introduction\tDomain adaptation is a machine learning paradigm that aims at improving the generalization performance of a new (target) domain by using a dataset from the original (source) domain. Suppose that, as the source domain dataset, we have a captioning corpus, consisting of images of daily lives and each image has captions. Suppose also that we would like to generate captions for exotic cuisine, which are rare in the corpus. It is usually very costly to make a new corpus for the target domain, i.e., taking and captioning those images. The research question here is how we can leverage the source domain dataset to improve the performance on the target domain. As described by Daum\u00e9 daume:07, there are mainly two settings of domain adaptation: fully supervised and semi-supervised. Our focus is the supervised setting, where both of the source and target domain datasets are labeled. We would like to use the label information of the source domain to improve the performance on the target domain. Recently, Recurrent Neural Networks (RNNs) have been successfully applied to various tasks in the field of natural language processing (NLP), including language modeling  , caption generation  and parsing  ."}
{"text": "System Design\tIn this section, we describe the system design, and workflow of SimplerVoice (Figure FIGREF1 ). SimplerVoice has 4 main components: input retrieval, object2text, text2visual, and output display. Figure FIGREF1 provides the overall structure of SimplerVoice system."}
{"text": "Analysis on the Cache-based Neural Model\tAs shown above, the topic cache is able to influence both the translations of beginning sentences and those of subsequent sentences while the dynamic cache built from translations of preceding sentences has an impact on the translations of subsequent sentences. We further study what roles the dynamic and topic cache play in the translation process. For this aim, we calculate the average number of words in translations generated by $RNNSearch*_{+Cd,Ct}$ that are also in the caches. During the counting process, stop words and \u201cUNK\u201d are removed from sentence and document translations. Table 4 shows the results. If only the topic cache is used ([ $document \\in [Ct]$ , $sentence \\in (Ct)$ ] in Table 4), the cache still can provide useful information to help NMT translate sentences and documents. 28.3 words per document and 2.39 words per sentence are from the topic cache."}
{"text": "A positive coefficient reflects an increased chance that an edge between two nodes with respective properties, like group affiliation, given all other parameters unchanged, is formed. Or, put differently, a positive coefficient implies that the probability of observing a network with a higher number of corresponding pairs relative to the hypothetical baseline network, is higher than to observe the baseline network itself  . For an intuitive interpretation, log-odds value of 0 corresponds to the even chance probability of INLINEFORM0 . Log-odds of INLINEFORM1 correspond to an increase of probability by INLINEFORM2 , whereas log-odds of INLINEFORM3 correspond to a decrease of probability by INLINEFORM4 . The computational challenges of estimating ERGMs is to a large degree due to the estimation of the normalizing constant. The number of possible networks is already extremely large for very small networks and the computation is simply not feasible. Therefore, an appropriate sample has to be found, ideally covering the most probable areas of the probability distribution. For this we make use of a method from the Markov Chain Monte Carlo (MCMC) family, namely the Metropolis-Hastings algorithm."}
{"text": "Experimental evaluation ::: Experimental settings\tWe conducted our experiments on the CMU ARCTIC database , which contains parallel recordings of professional US English speakers sampled at 16 kHz. One female (slt) was chosen as the target speaker and one male (bdl) and one female (clb) were chosen as sources. We selected 100 utterances each for validation and evaluation, and the other 932 utterances were used as training data. For the TTS corpus, we chose a US female English speaker (judy bieber) from the M-AILABS speech dataset  to train a single-speaker Transformer-TTS model. With the sampling rate also at 16 kHz, the training set contained 15,200 utterances, which were roughly 32 hours long. The entire implementation was carried out on the open-source ESPnet toolkit , , including feature extraction, training and benchmarking. We extracted 80-dimensional mel spectrograms with 1024 FFT points and a 256 point frame shift. The base settings for the TTS model and training follow the Transformer.v1 configuration in , and we made minimal modifications to it for VC."}
{"text": "These filters rule out around 95% of the raw data and we end up with 21.4M news articles, 12,000 of which are randomly sampled for validation. We pretrain the model for 10 epochs and evaluate its performance on the validation set at the end of each epoch. The model with the highest ROUGE-L score is selected. During pretraining, we use a dropout rate of 0.3 for all inputs to transformer layers. The batch size is 1,920. We use RAdam  as the optimizer, with a learning rate of $10^{-4}$. Also, due to the different numerical scales of the positional embedding and initialized sentence piece embeddings, we divide the positional embedding by 100 before feeding it into the transformer. The beam width is set to 5 during inference. Finetuning. During finetuning, we keep the optimizer, learning rate and dropout rate unchanged as in pretraining. The batch size is 32 for all datasets. We pick the model with the highest ROUGE-L score on the validation set and report its performance on the test set. Our strategy of Pretraining with unlabeled Lead-3 summaries is called PL."}
{"text": "Experiments\tWe introduce our setups of the CommonsenseQA dataset  , present the baseline methods, and finally analyze experimental results."}
{"text": "Conclusion\tIn this paper, we introduced the problem of conversational skill discovery in speech-based assistants and presented an approach to enable users to find skills. To this end, we started with a rule-based agent and improved it by using RL, enabling the agent to adapt to different user attributes and conversational styles. We compared popularity based, rule-based and RL-based model conversational agents by deploying them in a real production setting and showed that the RL agent learns to adapt its policy to achieve a higher success rate with shorter dialogs. For future work, we plan to further personalize the dialog policy based on user attributes and conversational preferences, and investigate richer state representations. Furthermore, we plan to explore the impact of evolving attributes and preferences on the learned policies."}
{"text": "Experiments ::: Baselines\tFor comparison, we choose two state-of-the-art baselines and their variants. E2ECM : In dialogue policy maker, it adopts a classic classification for skeletal sentence template. In our implement, we construct multiple binary classifications for each act to search the sentence template according to the work proposed by . CDM : This approach designs a group of classifications (two multi-class classifications and some binary classifications) to model the dialogue policy. E2ECM+RL: It fine tunes the classification parameters of the dialogue policy by REINFORCE . CDM+RL: It fine tunes the classification of the act and corresponding parameters by REINFORCE . In order to verify the performance of the dialogue policy maker, the utterance encoder and dialogue belief tracker of our proposed model and baselines is the same, only dialogue policy maker is different."}
{"text": "We use three tags $\\mathbf {B}_+$, $\\mathbf {E}_{S,+}$ and $\\mathbf {A}_+$ to encode such information above. The second sentiment span with positive ($+$) polarity contains a two-word target \u201cShin Lim\u201d. The word \u201cand\u201d appearing before such target takes a tag $\\mathbf {B}_+$. The words \u201cperform amazing magic\u201d appearing after such target take a tag $\\mathbf {A}_+$ at each position. As for the target, the word \u201cShin\u201d at the beginning of the target takes tags $\\mathbf {B}_+$ and $\\mathbf {E}_{B,+}$, while the word \u201cLim\u201d at the end of the target takes tags $\\mathbf {E}_{E,+}$ and $\\mathbf {A}_+$. The third sentiment span with neutral (0) polarity contains a single-word target \u201cAGT\u201d. Similarly, we use three tags $\\mathbf {B}_0$, $\\mathbf {E}_{S,0}$ and $\\mathbf {A}_0$ to represent such single word target. The word \u201con\u201d appearing before such target takes a tag $\\mathbf {B}_0$."}
{"text": "Prediction of these dialogue acts in a real-time setting can be leveraged to generate appropriate automated agent responses to similar situations. Additionally, important patterns can emerge from analysis of the fine-grained acts in a dialogue in a post-prediction setting. For example, if an agent does not follow-up with certain actions in response to a customer's question dialogue act, this could be found to be a violation of a best practice pattern. By analyzing large numbers of dialogue act sequences correlated with specific outcomes, various rules can be derived, i.e. \"Continuing to request information late in a conversation often leads to customer dissatisfaction.\" This can then be codified into a best practice pattern rules for automated systems, such as \"A request for information act should be issued early in a conversation, followed by an answer, informative statement, or apology towards the end of the conversation.\" In this work, we are motivated to predict the dialogue acts in conversations with the intent of identifying problem spots that can be addressed in real-time, and to allow for post-conversation analysis to derive rules about conversation outcomes indicating successful/unsuccessful interactions, namely, customer satisfaction, customer frustration, and problem resolution."}
{"text": "Various approaches, analogies, and metaphors have been taken in this area, including \u201cdatasheets for datasets\u201d , \u201dmodel cards\u201d , \u201cdata statements\u201d , \u201cnutrition labels\u201d , a \u201cbill of materials\u201d , \u201cdata labels\u201d , and \u201csupplier declarations of conformity\u201d . Many go far beyond the concerns we have raised around human-labeled training data, as some are also (or primarily) concerned with documenting other forms of training data, model performance and accuracy, bias, considerations of ethics and potential impacts, and more. We discuss how our findings relate to this broader emerging area more in the concluding discussion."}
{"text": "Evaluation\tIn this section, we present the collated results of our experiments. To show that our model achieves higher accuracy than the baseline model and on a smaller training dataset, we first fix the test dataset. Our test dataset, as mentioned before, consists of 5000 tweets. We conducted our experiments on an Intel Core i5 machine (4 cores), with 8 GB RAM. The following are the accuracies of the baseline model on a training set of 200K tweets:  We filtered the training set with a subjectivity threshold of 0.5. By doing this, we saw that the number of tweets reduced to approximately 0.6 million tweets from an earlier total of 1.6 million. We then trained our model described in earlier sections on a 100K tweets randomly picked from this filtered training dataset, and observed the following accuracies:  Note that all the accuracies in the tables above have been recorded as the average of 3 iterations of our experiment. We achieve higher accuracy for all feature vectors, on all classifiers, and that too from a training dataset half the size of the baseline one. We now see the intricacies of the subjectivity threshold parameter."}
{"text": "Experiments ::: Experimental Setups ::: Knowledge Distillation\tPrevious work on non-autoregressive generation ,  has used translations produced by a pre-trained autoregressive NMT model as the training data, noting that this can significantly improve the performance. We analyze the impact of distillation in \u00a7 SECREF45."}
{"text": "We however decided to keep them in the final corpus, as they could be useful in specific tasks and are easy to filter out otherwise. Besides the normalization described in SECREF13, no cleaning nor post-processing is applied to the sentences. This is a deliberate choice to avoid losing any information that could be pertinent for a given task or for further selection. As a result, the mean letter density is 80% and only 61% of sentences both start with an uppercase letter and end with a common punctuation mark (.!?). Finally, although we performed no human validation per se, we actively monitored the crawling process to spot problematic domains early. This allowed to blacklist some domains entirely, for example those serving embedded PDFs (impossible to parse properly) or written in very close German dialects."}
{"text": "For the final classification, an additional linear layer is added to output predicted class distributions. For noise handling, we experiment with the Confusion Matrix model by  and the Cleaning model by . We repeat all the Bi-LSTM experiments 20 times and report the average F1-score (following the approach by ) and the standard error. The BERT model is obtained by fine-tuning the pre-trained BERT embeddings on NER data with an additional untrained CRF classifier. We fine-tuned all the parameters of BERT including that of the CRF end-to-end. This has been shown to give better performance than using word features extracted from BERT to train a classifier . The evaluation result is obtained as an average of 5 runs, we report the F1-score and the standard error in the result section."}
{"text": "Finally, the difference between the best results and the results using the Local features is lower than for English, even though the Local results are similar to those obtained with the English datasets (except for Turkish, but this is due to the significantly smaller size of the data, as shown in Table TABREF7 ). We believe that all these four issues are caused, at least partially, by the lack of domain-specific clustering features used for the multilingual experiments. In other words, while for the English experiments we leveraged the Yelp dataset to train the clustering algorithms, in the multilingual setting we first tried with already available clusters induced from the Wikipedia. Thus, it is to be expected that the gains obtained by clustering features obtained from domain-specific data such as Yelp would be superior to those achieved by the clusters trained on out-of-domain data. In spite of this, Table TABREF20 shows that our system outperforms the best previous approaches across the five languages. In some cases, such as Turkish and Russian, the best previous scores were the baselines provided by the ABSA organizers, but for Dutch, French and Spanish our system is significantly better than current state-of-the-art."}
{"text": "Another advantage of social networks is that they are widely used. Twitter claims that they had 313 million monthly active users in the second quarter of 2016 . This characteristic will enlarge the scope of detection and is extremely helpful when dealing with cross domain attacks because tweets from multiple places can be leveraged. The large number of users of social networks will also guarantee the sensitivity of the model. However, because of the large number of users, a huge quantity of tweets will be generated in a short time, making it difficult to manually annotate the tweets, which makes unsupervised or weakly-supervised models much more desirable. In the Twitter data that we collected there are three kinds of tweets. Firstly are tweets that are actually about a cyberattack. For example, someone tweeted \"Can't sign into my account for bank of America after hackers infiltrated some accounts.\" on September 19, 2012 when a attack on the website happened. Secondly are tweets about some random complaints about an entity like \"Death to Bank of America!!!! RIP my Hello Kitty card... \" which also appeared on that day. Lastly are tweets about other things related to the bank."}
{"text": "The second point follows given that words embed on a manifold, sentences traces out paths on the manifold, and the difference equation the RNN solves bears a striking resemble to the first order equation for parallel transport, DISPLAYFORM0   where INLINEFORM0 is the INLINEFORM1 -th hidden state encountered when reading over a sentence and INLINEFORM2 is the RNN conditioned by the INLINEFORM3 -th word, INLINEFORM4 , acting on the hidden state. Since sentences trace out a path on the word manifold, and parallel transport operators for representations of the word manifold take values in the group, the RNN must parallel transport hidden states either on the group itself or on a base space, INLINEFORM5 , equipped with some word field, INLINEFORM6 , that connects the path in the base space to the path on the word manifold. Leveraging these results, we propose two new technologies. First, we propose a class of recurrent-like neural networks for NLP tasks that satisfy the differential equation DISPLAYFORM0  where DISPLAYFORM0   and where INLINEFORM0 and INLINEFORM1 are learned functions. INLINEFORM2 corresponds to traditional RNNs, with INLINEFORM3 ."}
{"text": "Experiments ::: Analysis on Decoding Speed ::: How does sentence length affect the decoding speed?\tNext, we examine if sentence length is a major factor affecting the decoding speed. We bucket the test data by the target sentence length. From Fig. FIGREF44, we can see that as the sentence length increases, FlowSeq achieves almost constant decoding time while Transformer has a linearly increasing decoding time. The relative decoding speed up of FlowSeq versus Transformer linearly increases as the sequence length increases. The potential of decoding long sequences with constant time is an attractive property of FlowSeq."}
{"text": "Proposed Model ::: Overall Architecture\tThe process of KANE is illustrated in Figure FIGREF2. We introduce the architecture of KANE from left to right. As shown in Figure FIGREF2, the whole triples of knowledge graph as input. The task of attribute embedding lays is embedding every value in attribute triples into a continuous vector space while preserving the semantic information. To capture both high-order structural information of KGs, we used an attention-based embedding propagation method. This method can recursively propagate the embeddings of entities from an entity's neighbors, and aggregate the neighbors with different weights. The final embedding of entities, relations and values are feed into two different deep neural network for two different tasks including link predication and entity classification."}
{"text": "Method ::: Hierarchically-Refined Review Encoder ::: Attention Inference Layer\tIn the attention inference layer, we model the dependencies between the original review and the summary with multi-head dot-product attention.Each head produces an attention matrix $\\mathbf {\\alpha } \\in \\mathbb {R}^{d_h \\times 1}$ consisting of a set of similarity scores between the hidden state of each token of the review text and the summary representation. The hidden state outputs are calculated by where $\\mathbf {W}_i^Q \\in \\mathbb {R}^{d_{h} \\times \\frac{d_{h}}{k}}$, $\\mathbf {W}_i^K \\in \\mathbb {R}^{d_{h} \\times \\frac{d_{h}}{k}}$ and $\\mathbf {W}_i^V \\in \\mathbb {R}^{d_{h} \\times \\frac{d_{h}}{k}}$ are model parameters. $Q$, $K$ and $V$ represent Query, Key and Value, respectively. $k$ is the number of parallel heads and $i \\in [1,k]$ indicates which head is being processed."}
{"text": "Standard Yor\u00f9b\u00e1 has 25 letters without the Latin characters c, q, v, x and z. There are 18 consonants (b, d, f, g, gb, j[dz], k, l, m, n, p[kp], r, s, \u1e63, t, w y[j]), 7 oral vowels (a, e, \u1eb9, i, o, \u1ecd, u), five nasal vowels, (an, $ \\underaccent{\\dot{}}{e}$n, in, $ \\underaccent{\\dot{}}{o}$n, un) and syllabic nasals (m\u0300, \u1e3f, \u01f9, \u0144). Yor\u00f9b\u00e1 is a tone language which makes heavy use of lexical tones which are indicated by the use of diacritics. There are three tones in Yor\u00f9b\u00e1 namely low, mid and high which are represented as grave ($\\setminus $), macron ($-$) and acute ($/$) symbols respectively. These tones are applied on vowels and syllabic nasals."}
{"text": "Dictionary-based PIE Extraction ::: String-based Extraction Methods ::: Additional Steps\tFor all string match-based methods, ways to improve performance are implemented, to make them as competitive as possible. Rather than doing exact string matching, we also allow words to be separated by something other than spaces, e.g. nuts-and-bolts for nuts and bolts. Additionally, there is an option to take into account case distinctions. With the case-sensitive option, case is preserved in the idiom lists, e.g. coals to Newcastle, and the string matching is done in a case-sensitive manner. This increases precision, e.g. by avoiding PIEs as part of proper names, but also comes at a cost of recall, e.g. for sentence-initial PIEs. Thirdly, there is the option to allow for a certain number of intervening words between each pair of words in the PIE. This should improve recall, at the cost of precision. For example, this would yield the true positive make a huge mountain out of a molehill for make a mountain out of a molehill, but also false positives like have a smoke and go for have a go."}
{"text": "Learning from Easy to Complex: Adaptive Multi-curricula Learning for Neural Dialogue Generation\tCurrent state-of-the-art neural dialogue systems are mainly data-driven and are trained on human-generated responses. However, due to the subjectivity and open-ended nature of human conversations, the complexity of training dialogues varies greatly. The noise and uneven complexity of query-response pairs impede the learning efficiency and effects of the neural dialogue generation models. What is more, so far, there are no unified dialogue complexity measurements, and the dialogue complexity embodies multiple aspects of attributes---specificity, repetitiveness, relevance, etc. Inspired by human behaviors of learning to converse, where children learn from easy dialogues to complex ones and dynamically adjust their learning progress, in this paper, we first analyze five dialogue attributes to measure the dialogue complexity in multiple perspectives on three publicly available corpora. Then, we propose an adaptive multi-curricula learning framework to schedule a committee of the organized curricula. The framework is established upon the reinforcement learning paradigm, which automatically chooses different curricula at the evolving learning process according to the learning status of the neural dialogue generation model. Extensive experiments conducted on five state-of-the-art models demonstrate its learning efficiency and effectiveness with respect to 13 automatic evaluation metrics and human judgments."}
{"text": "Los diccionarios consisten en entradas de pares: POS$_\\Phi $ y una lista de palabras y signos asociados, formalmente POS$_\\Phi $ $\\rightarrow $ $l$(POS$_\\Phi )=(l_1,l_2,...,l_j)$. Se reemplaza aleatoriamente cada POS$_\\Phi $ por una palabra de $l$ que corresponda a la misma clase gramatical. Las etiquetas POS$_\\lambda $ ser\u00e1n reemplazadas por las palabras producidas por Word2vec $L(Q)$. Si ninguna de las palabras de $L(Q)$ tiene la forma sint\u00e1ctica exigida por POS$_\\lambda $, empleamos la biblioteca PATTERN para realizar conjugaciones o conversiones de g\u00e9nero y/o n\u00famero y reemplazar correctamente POS$_\\lambda $."}
{"text": "We then process every sentence by checking if it contains \u201cdie\u201d or \u201cdat\u201d, and if so, add a training example for every occurrence of this word in the sentence, where a single occurrence is masked. For the test set for example, this resulted in about 289k masked sentences. We then test two different approaches for solving this task on this dataset. The first approach is making the BERT models use their MLM task and guess which word should be filled in this spot, and check if it has more confidence in either \u201cdie\u201d or \u201cdat\u201d (by checking the first 2,048 guesses at most, as this seemed sufficiently large). This allows us to compare the zero-shot BERT models, i.e. without any fine-tuning after pre-training, for which the results can be seen in Table TABREF7. The second approach uses the same data, but creates two sentences by filling in the mask with both \u201cdie\u201d and \u201cdat\u201d, appending both with the [SEP] token and making the model predict which of the two sentences is correct. The fine-tuning was performed using 4 Nvidia GTX 1080 Ti GPUs and evaluated against the same test set of 399k utterances."}
{"text": "For the SVMs in the generalization-dataset experiments, we tuned the hyperparameters on novel-9 with a grid search, and found the optimal setting as: stopping condition tol is 1e-5, at a max-iteration of 1,500."}
{"text": "Related Work\tWork toward learning reusable sentence encoders can be traced back at least as far as the multitask model of  , but has seen a recent surge in progress with the successes of CoVe  , ULMFit  , ELMo  , and the Transformer LM  . However, each uses a different model and dataset from the others, so while these works serve as existence proofs that effective reusable sentence encoders are possible, they do not address the question of what task or tasks should be used to create them. The revival of interest in sentence encoder pretraining is recent enough that relatively little has been done to understand the relative merits of these models, though two exceptions stand out. In unpublished work,  offer an analysis of the relative strengths of translation and language modeling using a single architecture and training dataset. They find that encoders trained as language models reliably uncover the most syntactic structure, even when they are trained on a strict subset of the data used for a comparable translation model.  offer a deeper investigation of model design issues for ELMo, showing that all of the standard architectures for sentence encoding can be effectively pretrained with broadly similar performance, and that all learn reasonably good representations of the morphological and syntactic properties of sentences."}
{"text": "Experiments ::: Variance reduction\tAs stated in Sec. SECREF3, the motivation of using the average reward baseline is for better variance reduction. Here we show it indeed is better in practice. The gradient variance is calculated as follows. At the end of each epoch, we take the saved model and run through the training set. We get the gradients from each training batch and calculate the variance for each parameter gradient across batches. To get a single value, we take the average of all the parameters. A mathematic expression of this process is: where $i$ is the index of each parameter; $b$ is the index of each batch; $\\theta $ is the network parameters; $\\text{grad}_{\\theta _i}^b$ is the gradient of $\\theta _i$ at batch $b$. As shown in Fig. FIGREF16, our method is always getting lower variance than SCST."}
{"text": "E\\left[ Y_{j} | \\mathbf {w}_{j}, \\alpha , \\phi ,\\eta , \\sigma ^{2} \\right] \\approx E_{q} \\left[\\mu \\left(\\eta ^{T} \\bar{\\mathbf {z}}_{j} \\right)\\right],"}
{"text": "Datasets and Evaluation Metrics\tWe use the DDI corpus of the 2013 DDIExtraction challenge  to train and test our model. The DDIs in this corpus are classified as five types. We give the definitions of these types and their example sentences, as shown in table TABREF4 . This standard dataset is made up of training set and testing set. We use the same metrics as in other drug-drug interaction extraction literature  ,  ,  ,  ,  ,  : the overall precision, recall, and F1 score on testing set. INLINEFORM0 denotes the set of {False, Mechanism, Effect, Advise, Int}. The precision and recall of each INLINEFORM1 are calculated by DISPLAYFORM0 DISPLAYFORM1  Then the overall precision, recall, and F1 score are calculated by DISPLAYFORM0  Besides, we evaluate the captured feature vectors with t-SNE  , a visualizing and intuitive way to map a high dimensional vector into a 2 or 3-dimensional space. If the points in a low dimensional space are easy to be split, the feature vectors are believed to be more distinguishable."}
{"text": "New Terminology: Potentially Idiomatic Expression (PIE)\tThe ambiguity of phrases like wake up and smell the coffee poses a terminological problem. Usually, these phrases are called idiomatic expressions, which is suitable when they are used in an idiomatic sense, but not so much when they are used in a literal sense. Therefore, we propose a new term: potentially idiomatic expressions, or PIEs for short. The term potentially idiomatic expression refers to those expressions which can have an idiomatic meaning, regardless of whether they actually have that meaning in a given context. So, see the light is a PIE in both `After another explanation, I finally saw the light' and `I saw the light of the sun through the trees', while it is an idiomatic expression in the first context, and a literal phrase in the latter context. The processing of PIEs involves three main challenges: the discovery of (new) PIE types, the extraction of instances of known PIE types in text, and the disambiguation of PIE instances in context. Here, we propose calling the discovery task simply PIE discovery, the extraction task simply PIE extraction, and the disambiguation task PIE disambiguation."}
{"text": "Question Aware Context Encoder\tThe question aware context embeddings $\\mathbf {C}$ is generated by the context encoder, while HAS-QA do not limit the use of context encoder. We choose a simple but efficient context encoder in this paper. It takes advantage of previous works  ,  , which contains the character-level embedding enhancement, the bi-directional attention mechanism  and the self-attention mechanism  . We briefly describe the process below . Word Embeddings: use size 300 pre-trained GloVe  word embeddings. Char Embeddings: encode characters in size 20, which are learnable. Then obtain the embedding of each word by convolutional layer and max pooling layer. Context Embeddings: concatenate word embeddings and char embeddings, and apply bi-directional GRU  to obtain the context embeddings. Both question and paragraph get their own context embeddings. Question Aware Context Embeddings: use bi-directional attention mechanism from the BiDAF  to build question aware context embeddings. Additionally, we subsequently apply a layer of self-attention to get the final question aware context embeddings."}
{"text": "Evaluation ::: Automatic Metrics\tFor automatically evaluating our methods, we propose to use widely used metric for image/video captioning. This is because the proposed CommonGen task can be regarded as also a caption task where the context are incomplete scenes with given concept-sets. Therefore, we choose BLEU-3/4 , ROUGE-2/L , CIDEr , and SPICE  as the main metrics. Apart from these classic metrics, we also include a novel embedding-based metric named BERTScore . To make the comparisons more clear, we show the delta of BERTScore results by subtracting the score of merely using input concept-sets as target sentences, named $\\triangle $BERTS. To have an estimation about human performance in each metric, we iteratively treat every reference sentence in dev/test data as the prediction to be compared with all references (including itself). That is, if a model has the same reasoning ability with average performance of our crowd workers, its results should exceed this \u201chuman bound\u201d."}
{"text": "Twitter Sentiment Analysis via Bi-sense Emoji Embedding and Attention-based LSTM\tSentiment analysis on large-scale social media data is important to bridge the gaps between social media contents and real world activities including political election prediction, individual and public emotional status monitoring and analysis, and so on. Although textual sentiment analysis has been well studied based on platforms such as Twitter and Instagram, analysis of the role of extensive emoji uses in sentiment analysis remains light. In this paper, we propose a novel scheme for Twitter sentiment analysis with extra attention on emojis. We first learn bi-sense emoji embeddings under positive and negative sentimental tweets individually, and then train a sentiment classifier by attending on these bi-sense emoji embeddings with an attention-based long short-term memory network (LSTM). Our experiments show that the bi-sense embedding is effective for extracting sentiment-aware embeddings of emojis and outperforms the state-of-the-art models. We also visualize the attentions to show that the bi-sense emoji embedding provides better guidance on the attention mechanism to obtain a more robust understanding of the semantics and sentiments."}
{"text": "Specifically, dialog act $$ is defined as the combination of intent $$ and slot-value pairs $\\lbrace (s_i, v_i)\\rbrace ^P_{i=1}$: where $P$ is the number of pairs, which varies in different dialog acts. Intents are usually used to distinguish different types of system actions. Typical examples include inform, request, confirm, select Slot-value pairs indicate the category and content of the information to express in the utterance, respectively. The goal of NLG is to translate $$ into a natural language response $= [x_1, \\cdots , x_T]$, where $T$ is the sequence length. In Figure FIGREF2 (b), we show an example of the dialog act: $\\textit {\\texttt {confirm}~(name=Hilton, area=center)}$, and the corresponding natural language response is \u201cLet me confirm that you are searching for Hilton in the center area\u201d."}
{"text": "Conversational Skill Discovery ::: Problem Formulation\tConversational skill discovery, similar to other goal-oriented dialog systems, can be formalized as a Markov Decision Process (MDP) . An MDP is a tuple $<\\mathcal {S}, \\mathcal {A}, \\mathcal {P}, \\mathcal {R}, \\mathcal {\\gamma }>$, where $\\mathcal {S}$ is the state space, $\\mathcal {A}$ is the action space, $\\mathcal {P}$ is the transition probability function, $\\mathcal {R}$ is the reward function, and $\\mathcal {\\gamma }$ is the discount factor. In this framework, at each time step t, the agent observes state $s_t \\in \\mathcal {S}$ and selects action $a_t \\in \\mathcal {A}$ according to its policy ($\\pi : \\mathcal {S} \\rightarrow \\mathcal {A}$). After performing the selected action, the agent receives the next state $s_{t+1}$ and a scalar reward $r_t$. The trajectory restarts after the agent reaches a terminal state."}
{"text": "\\nonumber \\end{split}$$   (Eq. 7)  where past QA pairs $q_1, a_1, \\dots , q_{k-1}, a_{k-1}$ in $C$ are concatenated and separated by two special tokens [Q] and [A] and then concatenate with the current question $q_k$ as the left side of BERT and the right side is the review document."}
{"text": "We took a subset of the instances equally distributed with respect to their class from the training set (330 instances for each class) and test set (71 instances for each class). As shown in Table TABREF31 , our classifier was able to correctly predict the underrepresented classes much better, although the overall accuracy is lower (59.9%). The result of this analysis clearly indicates that class imbalance has a negative impact on the system performance."}
{"text": "Introduction\tConversational agents or dialogue systems development are gaining more attention from both industry and academia  ,  in the latest years. Some works tried to model them into domain-specific tasks such as customer service  ,  , and shopping assistance  . Other works design a multi-purpose agents such as SIRI, Amazon Alexa, and Google Assistance. This domain is a well-researched area in Human-Computer Interaction research community but still become a hot topic now. The main development focus right now is to have an intelligent and humanizing machine to have a better engagement when communicating with human  . Having a better engagement will lead to higher user satisfaction, which becomes the main objective from the industry perspective. In this study, we will only focus on textual conversational agent or chatbot, a conversational artificial intelligence which can conduct a textual communication with a human by exploiting several natural language processing techniques. There are several approaches used to build a chatbot, start by using a simple rule-based approach  ,  until more sophisticated one by using neural-based technique  ,  ."}
{"text": "Sentiment analysis is contextual mining of text which identifies and extracts subjective information in the source material and helping a business to understand the social sentiment of their brand, product or service while monitoring online conversations. To resolve the risks and fraud problems, it is important not only to analyze the social-network opinion, but also to scan the smart contract vulnerability detection."}
{"text": "Model components ::: Attention Mechanism\tAttention mechanism was first developed by , in their paper \u201cNeural Machine Translation by Jointly Learning to Align and Translate\u201d which takes in as a natural extension of their previous work on the sequence to sequence Encoder-Decoder model. Attention is proposed as a solution to mitigate the limitation of the Encoder-Decoder architecture which encodes the input sequence to one fixed length vector from which the output is decoded at each time step. This problem seems to be more of a issue when decoding long sequences. Attention is proposed as a singular method to both align and translate. Alignment is the problem in machine translation that seeks to find which parts of the input sequence are relevant to each word in the output, whereas translation is the process of using the relevant information to select the appropriate output."}
{"text": "Conclusions and applications\tWe proposed a new method for clustering of visits in health centers based on descriptions written by doctors. We validated this new method on a new large corpus of Polish medical records. For this corpus we identified medical concepts and created their embeddings with GloVe algorithm. The quality of the embeddings was measured by the specific analogy task designed specifically for this corpus. It turns out that analogies work well, what ensures that concept embeddings store some useful information. Clustering was performed on visits embedding created based on word embedding. Visual and numerical examination of derived clusters showed an interesting structure among visits. As we have shown obtained segments are linked with medical diagnosis even if the information about recommendations nor diagnosis were not used for the clustering. This additionally convinces that the identified structure is related to some subgroups of medical conditions. Obtained clustering can be used to assign new visits to already derived clusters. Based on descriptions of an interview or a description of patient examination we can identify similar visits and show corresponding recommendations."}
{"text": "The Overfitting-Resistant Classifier\tThe classifier that we use to make a decision based on these features is scikit-learn's LinearSVC, using the default parameters in scikit-learn 0.19.1. (In Section SECREF4 , we discuss other models and parameters that we tried.) This classifier makes effective use of the features while being simple enough to avoid some amount of overfitting. One aspect of the classifier that made a noticeable difference was the scaling of the features. We tried INLINEFORM0 and INLINEFORM1 -normalizing the columns of the input matrix, representing the values of each feature, and decided on INLINEFORM2 normalization. We took advantage of the design of our features and the asymmetry of the task as a way to further mitigate overfitting. All of the features were designed to identify a property that INLINEFORM0 has and INLINEFORM1 does not, as is the case for the discriminative examples, so they should all make a non-negative contribution to a feature being discriminative. We can inspect the coefficients of the features in the SVC's decision boundary."}
{"text": "Datasets ::: Persona-chat and Cornell-movie corpus\tWe use two datasets to train our chatbot model. The first one is the Persona-chat  which contains dialogues between two people with different profiles that are trying to know each other. It is complemented by the Cornell-movie dialogues dataset , which contains a collection of fictional conversations extracted from raw movie scripts. Persona-chat's sentences have a maximum of 15 words, making it easier to learn for machines and a total of 162,064 utterances over 10,907 dialogues. While Cornell-movie dataset contains 304,713 utterances over 220,579 conversational exchanges between 10,292 pairs of movie characters."}
{"text": "Acknowledgments\tThis work has received funding from the European Research Council (ERC), under the European Union's Horizon 2020 research and innovation programme (FASTPARSE, grant agreement No 714150), from the TELEPARES-UDC (FFI2014-51978-C2-2-R) and ANSWER-ASAP (TIN2017-85160-C2-1-R) projects from MINECO, and from Xunta de Galicia (ED431B 2017/01)."}
{"text": "If removing this word already changes the model's prediction \u2013 that is, $M(\\mathbf {x}_{w_i = \\texttt {[MASK]}}) \\ne y$ \u2013, we select a random rare synonym $\\hat{w}_i \\in S(w_i)$ and add $(\\mathbf {x}_{w_i = \\hat{w}_i}, y)$ to the test set. Otherwise, we repeat the above procedure; if the label still has not changed after masking up to 5 words, we discard the corresponding entry. All so-obtained test set entries $(\\mathbf {x}_{w_{i_1} = \\hat{w}_{i_1}, \\ldots , w_{i_k} = \\hat{w}_{i_k} }, y)$ have the following properties: If each $w_{i_j}$ is replaced by a [MASK] token, the entry is classified incorrectly by $M$. In other words, understanding the words $w_{i_j}$ is essential for $M$ to determine the correct label."}
{"text": "A Large-Scale Corpus for Conversation Disentanglement\tDisentangling conversations mixed together in a single stream of messages is a difficult task, made harder by the lack of large manually annotated datasets. We created a new dataset of 77,563 messages manually annotated with reply-structure graphs that both disentangle conversations and define internal conversation structure. Our dataset is 16 times larger than all previously released datasets combined, the first to include adjudication of annotation disagreements, and the first to include context. We use our data to re-examine prior work, in particular, finding that 80% of conversations in a widely used dialogue corpus are either missing messages or contain extra messages. Our manually-annotated data presents an opportunity to develop robust data-driven methods for conversation disentanglement, which will help advance dialogue research."}
{"text": "Our approach allows to create a small number of seed patterns in advance, then explore other patterns through process of grammar learning. A similar bootstrapping semi-automatic approach to ontology learning was developed in  , where the user validates lexicalizations of a particular relation to learn new instances, and in  , where the user validates newly identified terms, while in our approach the user validates grammar rules to learn the composition of whole sentences. A similar approach with combining DBpedia with Wikipedia for superised learning has been taken in  , however their focus is more on lexicalization of relations and classes."}
{"text": "Tree Kernels compute the number of common substructures between two trees $T_1$ and $T_2$ without explicitly considering the whole fragment space . Suppose the set $\\mathcal {F}=\\lbrace f_1,f_2, \\dots , f_{|\\mathcal {F}|} \\rbrace $ be the tree fragment space and $\\mathcal {X}_i(n)$ be an indicator function that is 1 if the $f_i$ rooted at node $n$ and equals to 0, otherwise. Now, tree kernel over $T_1$ and $T_2$ is defined as below  : where $N_{T_1}$ and $N_{T_2}$ are the set of nodes of $T_1$ and $T_2$, respectively and which shows the number of common fragments rooted in $n_1$ and $n_2$ nodes. Different tree kernels vary in their definition of $\\Delta $ function and fragment type. There are three important characterizations of fragment type  : SubTree, SubSet Tree and Partial Tree. A SubTree is defined by taking a node of a tree along with all its descendants."}
{"text": "Semantic parsing takes advantage of RNN architectures to parse MWPs directly into equations or expressions in a math-specific language , . RNNs have shown promising results, but they have had difficulties balancing parenthesis, and also, sometimes incorrectly choose numbers when generating equations. Rehman et al.  used POS tagging and classification of equation templates to produce systems of equations from third-grade level MWPs. Most recently, Sun et al.  used a Bi-Directional LSTM architecture for math word problems. Huang et al.  used a deep reinforcement learning model to achieve character placement in both seen and novel equation templates. Wang et al.  also used deep reinforcement learning."}
{"text": "We then use the gold standard from the annotation project to assess the performance of multiple neural classification models trained exclusively on Term Frequency \u2013 Inverse Document Frequency (TF-IDF) vectorized EHR data, described in section 4. To further improve the performance of our model, we incorporate domain-relevant MWEs identified using all in-house data."}
{"text": "We got the best quality with regard to the F1-score with 256 as the size of the hidden states of the LSTMs, 1024 as word embedding size, and a dropout of 30 %. For the batch size, we used 64. We optimized our single-task model trained on real data in the same manner as the single-task baseline, except that we used 64 epochs. In the multi-task learning approach, we trained both tasks for 10 epochs. We use for $m$ (the instance multiplicator of the synthetic dataset) such a value that the synthetic dataset has nearly the size of one-tenth of the out-of-domain dataset. Because of long training times, we were not able to optimize the hyperparameters. We chose 256 as the size of the hidden states of the LSTMs, 1024 as word embedding size, and 50 % for the dropout and were not able to run multiple runs. For $n$ (the number of instances that are trained simultaneously on the GPU), we chose 128 and for $t$ (number of groups after that the model weights are updated) we chose 11."}
{"text": "The approximation error becomes larger when the derivative is evaluated further from the critical point. In contrast, the automatic differentiation (in both modes) yields the exact result of 0."}
{"text": "The AQA-QR rewrites are those whose corresponding answers are evaluated as AQA TopHyp in Table 1 . These single rewrites alone outperform the original SearchQA queries by 2% on the test set. We analyze the top hypothesis instead of the final output of the full AQA agent to avoid confounding effects from the answer selection step. These rewrites look different from both the Base-NMT and the SearchQA ones. For the example above AQA-QR's top hypothesis is What is name gandhi gandhi influence wrote peace peace?. Surprisingly, 99.8% start with the prefix What is name. The second most frequent is What country is (81 times), followed by What is is (70) and What state (14). This is puzzling as it occurs in only 9 Base-NMT rewrites, and never in the original SearchQA questions. We speculate it might be related to the fact that virtually all answers involve names, of named entities (Micronesia) or generic concepts (pizza). AQA-QR's rewrites seem less fluent than both the SearchQA and the Base-MT counterparts."}
{"text": "Satisfaction aspect has three categories, including affect, ethics and behaviour, and accessibility. Affect is the most suitable assessment categories for EAC. This category asses several quality aspects such as, chatbots' ability to convey personality, give conversational cues, provide emotional information through tone, inflexion, and expressivity, entertain and/or enable the participant to enjoy the interaction and also read and respond to moods of human participant  . Ethic and behaviour category focuses on how a chatbot can protect and respect privacy  . Other quality aspects, including sensitivity to safety and social concerns and trustworthiness  . The last categories are accessibility, which the main quality aspect focus to assess the chatbot ability to detect meaning or intent and, also responds to social cues ."}
{"text": "Introduction ::: blackTraditional Learning Based Text-to-image Synthesis\tIn the early stages of research, text-to-image synthesis was mainly carried out through a search and supervised learning combined process , as shown in Figure FIGREF4. In order to connect text descriptions to images, one could use correlation between keywords (or keyphrase) & images that identifies informative and \u201cpicturable\u201d text units; then, these units would search for the most likely image parts conditioned on the text, eventually optimizing the picture layout conditioned on both the text and the image parts. Such methods often integrated multiple artificial intelligence key components, including natural language processing, computer vision, computer graphics, and machine learning. The major limitation of the traditional learning based text-to-image synthesis approaches is that they lack the ability to generate new image content; they can only change the characteristics of the given/training images. Alternatively, research in generative models has advanced significantly and delivers solutions to learn from training images and produce new visual content. For example, Attribute2Image  models each image as a composite of foreground and background. In addition, a layered generative model with disentangled latent variables is learned, using a variational auto-encoder, to generate visual content."}
{"text": "To investigate whether a BiLSTM can assist the model, we apply another BiLSTM to the three fusion representations in Formula 23 respectively and then put them together. The results are shown in the second column in Table 3 , which indicate that the BiLSTM does not help improve the performance of the models."}
{"text": "Conclusions\tZipf's law is one of the most universal statistics of natural languages. However, it may be too universal. While it may not strictly apply to sequences of independent random symbols with random spacings , it appears to describe random texts that conserve most of the correlations between successive symbols, as accurately as it describes texts written in real languages. Further, Heaps-Herdan law and the degree distribution of the adjacency network, appear to be consequences of Zipf's law, and are, thus, as universal. In this work we studied 91 texts in seven different languages, as well as random texts constructed by randomizing the spacings between words without altering the order of the letters in the text. We find that they are all well described by the universal laws. However, we also found that the distribution of clustering coefficients of the networks of each text appears to vary from one language to another, and to distinguish random texts from real languages. The nodes that vary the most among the distributions of $C(k)$ are those for which $C(k)$ is equal to 0 or 1."}
{"text": "Task Definition\tWe consider a shared channel in which a group of people are communicating by sending messages that are visible to everyone. We label this data with a graph in which messages are nodes and edges indicate that one message is a response to another. Each connected component is a conversation. Figure shows an example of two entangled conversations and their graph structure. It includes a message that receives multiple responses, when multiple people independently help BurgerMann, and the inverse, when the last message responds to multiple messages. We also see two of the users, delire and Seveas, simultaneously participating in two conversations. This multi-conversation participation is common. The example also shows two aspects of IRC we will refer to later. Directed messages, an informal practice in which a participant is named in the message. These cues are useful for understanding the discussion, but only around 48% of messages have them. System messages, which indicate actions like users entering the channel. These all start with ===, but not all messages starting with === are system messages, as shown by the second message in Figure ."}
{"text": "Ablation Study\tWe conducted another two groups of experiments to investigate the contribution of the three different types of information: narrative sequence, sentiment evolution and commonsense knowledge. First, we measure the accuracy of only using one type of information at a time and describe the result in Table 3 . When we use just one type of information, the performances are worse than when using all of the information, suggesting a single type of information is insufficient for story ending selection. We also measure the performance of our model by stripping one type of information at a time and display the results in Table 4 . We observe that by removing the narrative sequence information, the model performance decreases most significantly. We suspect this is because the narrative chain is the key element that differentiates a story from other types of writing. Therefore, removing narrative sequence information makes it difficult to predict the story ending. If we only use the narrative sequence information, the performance is 85.3%. When commonsense knowledge is added to the model on top of the narrative sequence information, the performance improves to 87.2% which is statistically significant. When sentiment evolution information is added, the model only improves to 87.6%."}
{"text": "Furthermore, $\\text{SRL}_{\\text{SAOKE}}$ and $\\text{HypeNet}$ methods suffer from the OOV problem, such as unfamiliar words/phrases like the person name and school name in the last case of Table 6 . In this situation they may fail to produce a reasonable result. Logician is able to cope with unfamiliar words/phrases by exploiting the context information using deep RNN network with the help of copy mechanism."}
{"text": "to conduct an adversarial attack on character-level NMT by employing differentiable string-edit operations. The method of generating adversarial examples was same in their previous work  . Compared with Belinkov et al.  , the authors demonstrated that black-box adversarial examples were much weaker than black-box ones in most cases. Iyyer et al.  crafted adversarial examples by the use of SCPNS they proposed. They designed this model for generating syntactically adversarial examples without decreasing the quality of the input semantics. The general process mainly relied on the encoder-decoder architecture of SCPNS. Given a sequence and a corresponding target syntax structure, the authors encoded them by a bidirectional LSTM model and decoded by LSTM model augmented with soft attention over encoded states  and the copy mechanism  . They then modified the inputs to the decoder, aiming at incorporating the target syntax structure to generate adversarial examples. The syntactically adversarial sentences not only could fool pre-trained models, but also improved the robustness of them to syntactic variation. The authors also used crowdsourced experiment to demonstrate the validity of the generated."}
{"text": "Experiments ::: Predicting Individual Stock Movements\tOther than predicting the S&P 500 index, we also investigate the effectiveness of our approach on the problem of individual stock prediction using the test set. We count the amounts of individual company related news events for each company by name matching, and select five well known companies with sufficient news, Apple, Citigroup, Boeing Company, Google and Wells Fargo from four different sectors, which is classified by the Global Industry Classification Standard. For each company, we prepare not only news events about itself, but also news events about the whole companies in the sector. We use company news, sector news and all financial news to predict individual stock price movements, respectively. The experimental results and news statistics are listed in Table TABREF30. The result of individual stock prediction by only using company news dramatically outperforms that of sector news and all news, which presents a negative correlation between total used amounts of news events and model performance. The main reason maybe that company-related news events can more directly affect the volatility of company shares, while sector news and all news contain many irrelevant news events, which would obstruct our ANRES model's learning the underlaying stock price movement trends."}
{"text": "The largest difference between the coalitions in the European Parliament and on Twitter is on the far-right, where we observe ENL and NI as isolated blocks. The results shown in Fig FIGREF44 quantify the extent to which communication in one social context (Twitter) can explain cooperation in another social context (co-voting in the European Parliament). A positive value indicates that the matching behavior in the retweet network is similar to the one in the co-voting network, specific for an individual policy area. On the other hand, a negative value implies a negative \u201ccorrelation\u201d between the retweeting and co-voting of MEPs in the two different contexts. The bars in Fig FIGREF44 correspond to the coefficients from the edge covariate terms of the ERGM, describing the relationship between the retweeting and co-voting behavior of MEPs. The coefficients are aggregated for individual policy areas by means of a meta-analysis. Overall, we observe a positive correlation between retweeting and co-voting, which is significantly different from zero. The strongest positive correlations are in the areas Area of freedom, security and justice, External relations of the Union, and Internal markets."}
{"text": "Materials and Methods\tWe describe and motivate the materials first and then proceed to the methods and motivations thereof. There are multiple ontologies available over internet with competency questions provided, but since the focus of our research is on SPARQL-OWL queries, we selected only those ontologies with CQs stated against ontology schema (T-Box). As a result we selected 5 ontologies with 234 competency questions in total. Table TABREF8 summarizes our dataset size and source of each ontology. The Software Ontology (SWO)  is included because its set of CQs is of substantial size and it was part of Ren et al.'s set of analysed CQs. The CQ sets of Dem@Care  and OntoDT  were included because they were available. CQs for the Stuff  and African Wildlife (AWO)  ontologies were added to the set, because the ontologies were developed by one of the authors (therewith facilitating in-depth domain analysis, if needed), they cover other topics, and are of a different `type' (a tutorial ontology (AWO) and a core ontology (Stuff)), thus contributing to maximising diversity in source selection."}
{"text": "31)   where the output embedding $W^2 \\in \\mathbb {R}^{d_\\mathrm {word} \\times V_\\mathrm {ext}}$ is tied with the corresponding part of the input embedding  , and $W^1 \\in \\mathbb {R}^{d_\\mathrm {word} \\times d}$ and $b^1 \\in \\mathbb {R}^{d_\\mathrm {word}}$ are learnable parameters. $P^v(y_t)$ is zero if $y_t$ is an out-of-vocabulary word for $V$ . The copy mechanism used in the original pointer-generator is based on the attention weights of a single-layer attentional RNN decoder  . The attention weights in our decoder stack are the intermediate outputs in multi-head attentions and are not suitable for the copy mechanism. Therefore, our model also uses additive attentions for the question and multiple passages on top of the decoder stack."}
{"text": "Results\tThe following sections give an overview of the results for all political bias prediction tasks. Some interpretations of the results are highlighted and a web application of the models is presented at the end of the section."}
{"text": "Conclusion and Outlook\tApplying the example of metal music, our work examined the textual topics found in song lyrics and investigated the association between these topics and high-level music features. By using LDA and MDS in order to explore prevalent topics and the topic space, typical text topics identified in qualitative analyses could be confirmed and objectified based on a large text corpus. These include e.g. satanism, dystopia or disgusting objects. It was shown that musical hardness is particularly associated with harsh topics like \u2018brutal death\u2019 and \u2018dystopia\u2019, while it is negatively linked to relatively mundane topics concerning personal life and love. We expect that even stronger correlations could be found for metal-specific topics when including more genres covering a wider range of hardness/darkness values. Therefore, we suggest transferring the method to a sample including multiple genres. Moreover, an integration with metadata such as genre information would allow for the testing of associations between topics, genres and high-level audio features. This could help to better understand the role of different domains in an overall perception of genre-defining attributes such as hardness."}
{"text": "Some of them require noise rate to be known a priori in order to tune their models during training , . Our approach combines direction (1) and direction (3), and trains two networks jointly without knowing the noise rate. We have used five latest existing methods in our experiments for SSC. The experimental results show that they are inferior to our proposed method. In addition, , , , , , and  studied weakly-supervised DNNs or semi-supervised DNNs. But they still need some clean-labeled training data. We use no clean-labeled data."}
{"text": "Introduction\tWith the rapid development in deep learning, there is a recent boom of task-oriented dialogue systems in terms of both algorithms and datasets. The goal of task-oriented dialogue is to fulfill a user's requests such as booking hotels via communication in natural language. Due to the complexity and ambiguity of human language, previous systems have included semantic decoding  to project natural language input into pre-defined dialogue states. These states are typically represented by slots and values: slots indicate the category of information and values specify the content of information. For instance, the user utterance \u201ccan you help me find the address of any hotel in the south side of the city\u201d can be decoded as $inform(area, south)$ and $request(address)$, meaning that the user has specified the value south for slot area and requested another slot address. Numerous methods have been put forward to decode a user's utterance into slot values. Some use hand-crafted features and domain-specific delexicalization methods to achieve strong performance , .  employs CNN and pretrained embeddings to further improve the state tracking accuracy.  extends this work by using two additional statistical update mechanisms.  uses human teaching and feedback to boost the state tracking performance."}
{"text": "in the form of multiple response generators), and we define a fixed time window for each major step in our pipeline, after which we interrupt the current component and use the information already computed from the sub-components in the next step, reducing total processing time. We will develop our system in three phases (Figure FIGREF1): Components marked core, core+, and core++ are to be completed by the end of phase 5, 7, and 9, respectively. These are the minimally planned components for each category, but if time permits we will advance faster. This provides us an incremental and iterative approach to build our architecture starting with the most important components, always testing included components before advancing to new ones. Our main novelties include: [topsep=2pt,itemsep=-1ex,partopsep=1ex,parsep=1ex] Using self-play strategies to train a neural response ranker. Computing a large number of metrics for both input and response, and specifically optimizing some models for our metrics. Training a separate dialog model for each user. Using a response classification predictor and a response classifier to predict and control aspects of responses. Predicting which model emits the best response before response generation."}
{"text": "While the original technique combines the word and sentence level attentions, our model combines the passage-level relevance $\\beta ^{p_k}$ and word-level attentions $\\alpha ^p_t$ by using simple scalar multiplication and re-normalization. The updated word attention is:  $$\\alpha ^p_{tl} & := \\frac{\\alpha ^p_{tl} \\beta ^{p_{k(l)} }}{\\sum _{l^{\\prime }} \\alpha ^p_{tl^{\\prime }} \\beta ^{p_{k(l^{\\prime })}}}.$$   (Eq. 39)"}
{"text": "Results of Sentiment Analysis of Twitter Messages\tIn this section, we provide a detailed discussion of the results obtained from the sentiment analysis of the two datasets. Fig. FIGREF11 corresponds to the histogram of sentiment polarities of tweets on Coronavirus by general public. It can be seen that majority of the tweets have a neutral sentiment followed by positive. The same can be inferred from Table TABREF10 that shows that around 54$\\%$ tweets are neutral, 29$\\%$ positive and a mere 15$\\%$ is negative. Fig. FIGREF12 corresponds to the histogram of sentiment polarities of tweets on Coronavirus by WHO. It can be seen that majority of the tweets have a neutral and positive sentiment. Table TABREF10 that shows that around 60$\\%$ tweets are positive, 24$\\%$ neutral and a mere 15$\\%$ is negative. This shows how WHO is trying to retain the positive spirit through its social media accounts. Fig. FIGREF13 and FIGREF14 represent the histograms produced by removing the neutral tweets. It readily reiterates that the positive emotions in the tweets are higher than negative ones."}
{"text": "\\end{split}$$   (Eq. 12)  where, $\\mathbf {S}$ is the permutation set of the reference representation and the inference representation. ${o}_{utn}^{(s^{\\prime })}$ and ${o}_{utn}^{(r)}$ is the frame level acoustic representation of permutation $s^{\\prime }$ and the reference clean speech, respectively. In each frame $t$ of the utterance $u$ , the overall minimum square error, $MSE$ , is obtained by comparing all the reference and inference representations of each permutation $s^{\\prime }$ . The architecture for pre-training the speaker tracing module is explained in Figure 2 (c)."}
{"text": "Reasoning Workflow\tAs shown in Fig. 2 , our framework accepts a pair of question and answer (QA-pair) denoted as $q$ and $a$ . It first recognizes the mentioned concepts within them respectively from the concept set $V$ of the knowledge graph. We then algorithmically construct the schema graph $g$ by finding paths between pairs of mentioned concepts (\u00a7 \"Schema Graph Grounding\" ). The grounded schema graph is further encoded with our proposed knowledge-aware graph network module (\u00a7 \"Knowledge-Aware Graph Network\" ). We first use a model-agnostic language encoder, which can either be trainable or a fixed feature extractor, to represent the QA-pair as a statement vector. The statement vector serves as an additional input to a GCN-LSTM-HPA architecture for path-based attentive graph modeling to obtain a graph vector. The graph vector is finally fed into a simple multi-layer perceptron to score this QA-pair into a scalar ranging from 0 to 1, representing the plausibility of the inference. The answer candidate with the maximum plausibility score to the same question becomes the final choice of our framework."}
{"text": "Results and Discussion ::: Performance on DROP's Development Set\tTable TABREF24 shows the results on DROP's development set. Compared to our base models, our large models exhibit a substantial improvement across all metrics."}
{"text": "Sentence Selection\tA summary is obliged to offer both informative and non-redundant content. While AttSum focuses on sentence ranking, it employs a simple greedy algorithm, similar to the MMR strategy  , to select summary sentences. At first, we discard sentences less than 8 words like the work of  . Then we sort the rest in descending order according to the derived ranking scores. Finally, we iteratively dequeue the top-ranked sentence, and append it to the current summary if it is non-redundant. A sentence is considered non-redundant if it contains significantly new bi-grams compared with the current summary content. We empirically set the cut-off of the new bi-gram ratio to 0.5."}
{"text": "Word-Level Syntactic Probes for Contextual Embeddings ::: Notation\tLet $S$ be a random variable ranging over all possible sequences of words. For the sake of this paper, we assume the vocabulary $\\mathcal {V}$ is finite and, thus, the values $S$ can take are in $\\mathcal {V}^*$. We write $\\mathbf {s}\\in S$ as $\\mathbf {s}= w_1 \\cdots w_{|\\mathbf {s}|}$ for a specific sentence, where each $w_i \\in \\mathcal {V}$ is a specific word in the sentence and the position $i \\in \\mathbb {N}^{+}$. We also define the random variable $W$ that ranges over the vocabulary $\\mathcal {V}$. We define both a sentence-level random variable $S$ and a word-level random variable $W$ since each will be useful in different contexts during our exposition."}
{"text": "Model Description\tAs shown in Figure FIGREF3 , our model contains two key components, namely Truncated History-Attention (THA) and Selective Transformation Network (STN), for capturing aspect detection history and opinion summary respectively. THA and STN are built on two LSTMs that generate the initial word representations for the primary ATE task and the auxiliary opinion detection task respectively. THA is designed to integrate the information of aspect detection history into the current aspect feature to generate a new history-aware aspect representation. STN first calculates a new opinion representation conditioned on the current aspect candidate. Then, we employ a bi-linear attention network to calculate the opinion summary as the weighted sum of the new opinion representations, according to their associations with the current aspect representation. Finally, the history-aware aspect representation and the opinion summary are concatenated as features for aspect prediction of the current time step. As Recurrent Neural Networks can record the sequential information  , we employ two vanilla LSTMs to build the initial token-level contextualized representations for sequence labeling of the ATE task and the auxiliary opinion word detection task respectively. For simplicity, let INLINEFORM0 denote an LSTM unit where INLINEFORM1 is the task indicator."}
{"text": "Semantic Decomposition\tIn this study, we use GloVe  as the source algorithm for learning dense word vectors. The entire content of English Wikipedia is utilized as the corpus. In the preprocessing step, all non-alphabetic characters (punctuations, digits, etc.) are removed from the corpus and all letters are converted to lowercase. Letters coming after apostrophes are taken as separate words (she'll becomes she ll). The resulting corpus is input to the GloVe algorithm. Window size is set to 15, vector length is chosen to be 300 and minimum occurrence count is set to 20 for the words in the corpus. Default values are used for the remaining parameters. The word embedding matrix, $\\mathcal {E}$ , is obtained from GloVe after limiting vocabulary to the most frequent 50,000 words in the corpus (i.e. $\\mathcal {E}$ is 50,000 $\\times $ 300). The GloVe algorithm is again used for the second time on the same corpus generating a second embedding space, $\\mathcal {E}^2$ , to examine the effects of different initializations of the word vectors prior to training."}
{"text": "Knowledge Graph Representation\tA knowledge graph $\\mathcal {G}$ comprises of set of facts represented as triplets ( $e^s,r,e^o$ ) denoting the relationship $r$ between subject entity $e^s$ and object entity $e^o$ . Associated to this knowledge graph, we have a set of attributes that describe observed characteristics of an entity. Attributes are represented as set of key-value pairs for each entity and an attribute can have null (missing) value for an entity. We follow Open World Assumption - triplets not observed in knowledge graph are considered to be missing but not false. We assume that there are no duplicate triplets or self-loops."}
{"text": "RL solvers have been used to find the optimal dialog policy (e.g.,  ;  ;  ;  ). In this context, at each turn the agent acts based on its understanding of what the user said, and reward function is modeled in terms of various dimensions of the interaction such as per-interaction user satisfaction, accomplishment of the task, efficiency of interaction, and dialog duration. Recently, deep RL has also been applied to the problem of dialog management and has shown improvements over rule-based systems , , , , . In this paper, we adopt the above formalism with the goal of training a dialog policy which allows the agent to take actions that maximize its success rate (measured in terms of number of dialogs which result in launching a skill) while providing a flexible and natural way for the user to navigate throughout various dialog states. In each turn of the dialog, the agent makes its decisions based on various available information such as user's intent (e.g., asking for a particular skill), the category the user has selected, whether the user is a first-time user, etc."}
{"text": "Data Collection ::: Dialogue Collection ::: User Side\tThe user state is the same as the user goal before a conversation starts. At each turn, the user needs to 1) modify the user state according to the system response at the preceding turn, 2) select some semantic tuples in the user state, which indicates the dialogue acts, and 3) compose the utterance according to the selected semantic tuples. In addition to filling the required values and updating cross-domain informable slots with real values in the user state, the user is encouraged to modify the constraints when there is no result under such constraints. The change will also be recorded in the user state. Once the goal is completed (all the values in the user state are filled), the user can terminate the dialogue."}
{"text": "Implicit schedules ::: Optimization details\tScaling either the gradients $g_t$ or the per-task learning rates $\\alpha $ is equivalent with standard stochastic gradient descent, but not with adaptive optimizers such as Adam , whose update rule is given in Eq. DISPLAY_FORM5. Moreover, sharing or not the optimizer accumulators (eg. running average of 1st and 2nd moment $\\hat{m}_t$ and $\\hat{v}_t$ of the gradients) is also impactful. Using separate optimizers and simultaneously scaling the gradients of individual tasks is ineffective. Indeed, Adam is scale-insensitive because the updates are divided by the square root of the second moment estimate $\\hat{v}_t$. The opposite scenario, a shared optimizer across tasks with scaled learning rates, is also problematic as the momentum effect ($\\hat{m}_t$) will blur all tasks together at every update. All experiments we present use distinct optimizers, with scaled learning rates. The converse, a shared optimizer with scaled gradients, could also potentially be employed."}
{"text": "Analysis of Preference\tRemember that we use a preference vector INLINEFORM0 to represent the user's preference over the columns in the knowledge base. Therefore, we investigate the learned arguments grouped by profile attributes. As seen in Figure FIGREF31 , the model successfully learns the fact that young people prefer social media as their contact information, while middle-aged and elderly people prefer phone number. The result shows great potential and advantage of end-to-end models. They are capable of learning meaningful intermediate arguments while being much simpler than existing reinforcement learning methods and pipeline models for the task of personalization in dialogs."}
{"text": "Introduction\tSocial media has become a popular medium for individuals to express opinions and concerns on issues impacting their lives  ,  ,  . In countries without adequate internet infrastructure, like Uganda, communities often use phone-in talk shows on local radio stations for the same purpose. In an ongoing project by the United Nations (UN), radio-browsing systems have been developed to monitor such radio shows  ,  . These systems are actively and successfully supporting UN relief and developmental programmes. The development of such systems, however, remains dependent on the availability of transcribed speech in the target languages. This dependence has proved to be a key impediment to the rapid deployment of radio-browsing systems in new languages, since skilled annotators proficient in the target languages are hard to find, especially in crisis conditions. In a conventional keyword spotting system, where the goal is to search through a speech collection for a specified set of keywords, automatic speech recognition (ASR) is typically used to generate lattices which are then searched to predict the presence or absence of keywords  ,  . State-of-the-art ASR, however, requires large amounts of transcribed speech audio  ,  ."}
{"text": "Model ::: Problem formulation\tWe formalize the problem of supervised abstractive summarization as follows. The input consists of $a$ pairs of articles and summaries: $\\lbrace (X_1, Y_1), (X_2, Y_2), ..., (X_a, Y_a)\\rbrace $. Each article and summary are tokenized: $X_i=(x_1,...,x_{L_i})$ and $Y_i=(y_1,...,y_{N_i})$. In abstractive summarization, the summary tokens need not be from the article. For simplicity, we will drop the data index subscript. The goal of the system is to generate summary $Y=(y_1,...,y_m)$ given the transcript $X=\\lbrace x_1, ..., x_n\\rbrace $."}
{"text": "t^{\\prime }) \\mid t^{\\prime } \\in \\mathcal {E}, (h, r, t^{\\prime }) \\notin \\mathcal {G} \\rbrace $ is the set of incorrect triples generated by corrupting a correct triple $(h, r, t)\\in \\mathcal {G}$ ."}
{"text": "We would compare the best performing training method to the state-of-the-art, and then do side-by-side comparison between them: Pre-large (P): To leverage the controllable attention mechanism for exploring the linguistic information in the large monolingual corpora, we create text only pre-training set by combining the first 10 million sentences of the WMT News Crawl datasets from 2007 to 2017 with 10 times M30k-half. This ends up in a large text only dataset of 10145000 unparalleled sentences in each language. P-txt: We would then pre-train our model without the image encoder on this dataset and use the M30k-half validation set for validation. P-txt-img: Once the text-only model is pre-trained, we then use it for the following fine-tuning stage on M30k-half. Except for the image encoder, we initialize our model with the pre-trained model parameters. The image encoder uses pre-trained ResNet-152  . The error gradient does not back-propagate to the original ResNet network. Scratch (S): We are also curious about the role of image can play when no pre-training is involved."}
{"text": "Another important aspect of the theory of definition is the distinction between essential and non-essential properties. As pointed by Burek burek2004adoption, stating that \u201ca human is an animal\u201d informs an essential property for a human (being an animal), but the sentence \u201chuman is civilized\u201d does not communicate a fundamental property, but rather something that happens to be true for humans, that is, an incidental property. Analyzing a subset of the WordNet definitions to investigate their structure, we noticed that most of them loosely adhere to the classical theory of definition: with the exception of some samples of what could be called ill-formed definitions, in general they are composed by a linguistic structure that resembles the genus-differentia pattern, plus optional and variable incidental properties. Based on this analysis, we derived a set of semantic roles representing the components of a lexical definition, which are described next."}
{"text": "Future work may benefit from replacing ELMo with other types of contextualized representations such as BERT in our scaffold model. For example, at the time of finalizing the camera ready version of this paper,  showed that a BERT contextualized representation model  trained on scientific text can achieve promising results on the SciCite dataset."}
{"text": "Our approach\tIn this section, we will give a detailed introduction to our CRU model. Firstly, we will give a brief introduction to GRU  as preliminaries, and then three variants of our CRU model will be illustrated."}
{"text": "Introduction\tAutomatically generating captions for images, namely image captioning , , has emerged as a prominent research problem at the intersection of computer vision (CV) and natural language processing (NLP). This task is challenging as it requires to first recognize the objects in the image, the relationships between them, and finally properly organize and describe them in natural language. Inspired by the sequence-to-sequence model for machine translation, most image captioning approaches adopt an encoder-decoder paradigm, which uses a deep convolutional neural network (CNN) to encode the input image as a vectorial representation, and a recurrent neural network (RNN) based caption decoder to generate the output caption. Recently, self-attention (SA) networks, denoted as SANs, have been introduced by ,  to replace conventional RNNs in image captioning. Since its first introduction in Transformer , SA and its variants have shown promising empirical results in a wide range of CV , , , , ,  and NLP , ,  tasks. Although SAN-based framework has achieved state-of-the-art performance in image captioning, it remains two problems to be solved. Firstly, SA is susceptible to the internal covariate shift  problem."}
{"text": "Bi-Directional Recurrent Neural Network\t Fig. 1 shows the simplest form of the BiRNN  . Unlike its uni-directional version, the BiRNN makes use of two recurrent states, one going in the forward direction in time $\\overrightarrow{\\mathbf {h}}_{t}$ and another in the backward direction $\\overleftarrow{\\mathbf {h}}_{t}$ to model past (history) and future information respectively. The past information can be modelled by  $$\\overrightarrow{\\mathbf {h}}_{t} = \\sigma (\\mathbf { W}^{(\\overrightarrow{{h}})}\\overrightarrow{\\mathbf {h}}_{t-1} + \\mathbf { W}^{(x)}\\mathbf {x}_{t})$$   (Eq. 4)  where $\\mathbf {x}_{t}$ is an input feature vector at time $t$ , $\\mathbf {W}^{(x)}$ is an input matrix, $\\mathbf {W}^{(\\overrightarrow{{h}})}$ is a history matrix and $\\sigma $ is an element-wise non-linearity such as a sigmoid."}
{"text": "The Taskmaster Corpus ::: Two-person, spoken dataset ::: WOz platform and data pipeline\tWhile it is beyond the scope of this work to describe the entire system in detail, there are several platform features that help illustrate how the process works.  Modality: The agents playing the assistant type their input which is in turn played to the user via text-to-speech (TTS) while the crowdsourced workers playing the user speak aloud to the assistant using their laptop and microphone. We use WebRTC to establish the audio channel. This setup creates a digital assistant-like communication style.  Conversation and user quality control: Once the task is completed, the agents tag each conversation as either successful or problematic depending on whether the session had technical glitches or user behavioral issues. We are also then able to root out problematic users based on this logging.  Agent quality control: Agents are required to login to the system which allows us to monitor performance including the number and length of each session as well as their averages.  User queuing: When there are more users trying to connect to the system than available agents, a queuing mechanism indicates their place in line and connects them automatically once they move to the front of the queue."}
{"text": "Introduction\tVirtual assistants help users accomplish tasks including but not limited to finding flights, booking restaurants and, more recently, navigating user interfaces, by providing a natural language interface to services and APIs on the web. The recent popularity of conversational interfaces and the advent of frameworks like Actions on Google and Alexa Skills, which allow developers to easily add support for new services, has resulted in a major increase in the number of application domains and individual services that assistants need to support, following the pattern of smartphone applications. Consequently, recent work has focused on scalable dialogue systems that can handle tasks across multiple application domains. Data-driven deep learning based approaches for multi-domain modeling have shown promise, both for end-to-end and modular systems involving dialogue state tracking and policy learning. This line of work has been facilitated by the release of multi-domain dialogue corpora such as MultiWOZ , M2M  and FRAMES . However, existing datasets for multi-domain task-oriented dialogue do not sufficiently capture a number of challenges that arise with scaling virtual assistants in production. These assistants need to support a large , constantly increasing number of services over a large number of domains. In comparison, existing public datasets cover few domains."}
{"text": "Multilingual Twitter Corpus and Baselines for Evaluating Demographic Bias in Hate Speech Recognition\tExisting research on fairness evaluation of document classification models mainly uses synthetic monolingual data without ground truth for author demographic attributes. In this work, we assemble and publish a multilingual Twitter corpus for the task of hate speech detection with inferred four author demographic factors: age, country, gender and race/ethnicity. The corpus covers five languages: English, Italian, Polish, Portuguese and Spanish. We evaluate the inferred demographic labels with a crowdsourcing platform, Figure Eight. To examine factors that can cause biases, we take an empirical analysis of demographic predictability on the English corpus. We measure the performance of four popular document classifiers and evaluate the fairness and bias of the baseline classifiers on the author-level demographic attributes."}
{"text": "In Danish this could be a post such as Kalle er faggot... In sub-task B the goal is to classify the type of offensive language by determining if the offensive language is targeted or not. Targeted offensive language contains insults and threats to an individual, group, or others  . Untargeted posts contain general profanity while not clearly targeting anyone  . Only posts labeled as offensive (OFF) in sub-task A are considered in this task. Each sample is annotated with one of the following labels: Targeted Insult (TIN). In English this could be a post such as @USER Please ban this cheating scum. In Danish this could be e.g. Hun skal da selv have 99 \u00e5r, den smatso. Untargeted (UNT). In English this could be a post such as 2 weeks of resp done and I still don't know shit my ass still on vacation mode. In Danish this could e.g. Dumme svin... In sub-task C the goal is to classify the target of the offensive language. Only posts labeled as targeted insults (TIN) in sub-task B are considered in this task  ."}
{"text": "While we implemented both an update-incremental and a restart-incremental RASA nlu component, the results of the two cannot be directly compared for accuracy as the underlying models differ greatly (i.e., sium is generative, whereas Tensorflow Embedding is a discriminative neural network; moreover, sium was designed to work as a reference resolution component to physical objects, not abstract intents), nor are these results conducive to an argument of update- vs. restart-incremental approaches, as the underlying architecture of the models vary greatly."}
{"text": "The Visual STS Dataset\tSTS assesses the degree to which two sentences are semantically equivalent to each other. The annotators measure the similarity among sentences, with higher scores for more similar sentences. The annotations of similarity were guided by the scale in Table TABREF4, ranging from 0 for no meaning overlap to 5 for meaning equivalence. Intermediate values reflect interpretable levels of partial overlap in meaning. In this work, we extend the STS task with images, providing visual information that models use, and assess how much visual content can contribute in a language understanding task. The input of the task now consists of two items, each comprising an image and its corresponding caption. In the same way as in STS, systems need to score the similarity of the sentences with the help of the images. Figure FIGREF1 shows an example of an instance in the dataset. In previous work reported in a non-archival workshop paper , we presented a preliminary dataset which used the text-only ground-truth similarity scores."}
{"text": "(Apple CEO Tim Cook introduces 2 new, larger iPhones, Smart Watch at Cupertino Flint Center event) The algorithm will output: Output: vietnam\u27e8O\u27e9Gi\u00e1m \u0111\u1ed1c \u0111i\u1ec1u h\u00e0nh\u27e8O\u27e9 \u27e8PER\u27e9Tim Cook\u27e8PER\u27e9 \u27e8O\u27e9c\u1ee7a\u27e8O\u27e9 \u27e8ORG\u27e9Apple\u27e8ORG\u27e9 \u27e8O\u27e9v\u1eeba gi\u1edbi thi\u1ec7u 2 \u0111i\u1ec7n tho\u1ea1i iPhone, \u0111\u1ed3ng h\u1ed3 th\u00f4ng minh m\u1edbi, l\u1edbn h\u01a1n \u1edf s\u1ef1 ki\u1ec7n\u27e8O\u27e9 \u27e8ORG\u27e9Flint Center\u27e8ORG\u27e9, \u27e8LOC\u27e9Cupertino\u27e8LOC\u27e9. With LOC, PER, ORG is Name of location, person, organization respectively. Note that O means Other (Not a Name entity). We will not denote the O label in the following examples in this article because we only care about name of entities. In this paper, we analyze common errors of the previous state-of-the-art techniques using Deep Neural Network (DNN) on VLSP Corpus. This may contribute to the later researchers the common errors from the results of these state-of-the-art models, then they can rely on to improve the model."}
{"text": "Our definition of bias loosely follows the idea of causal testing: matched pairs of individuals (instances) that differ in only a targeted concept (like gender) are evaluated by a model and the difference in outcomes (or scores) is interpreted as the causal influence of the concept in the scrutinized model. The definition is parametric in the scoring function and the target concept. Natural scoring functions exist for a number of neural natural language processing tasks. We instantiate the definition for two important tasks\u2014coreference resolution and language modeling. Coreference resolution is the task of finding words and expressions referring to the same entity in a natural language text. The goal of language modeling is to model the distribution of word sequences. For neural coreference resolution models, we measure the gender coreference score disparity between gender-neutral words and gendered words like the disparity between \u201cdoctor\u201d and \u201che\u201d relative to \u201cdoctor\u201d and \u201cshe\u201d pictured as edge weights in Figure FIGREF2 . For language models, we measure the disparities of emission log-likelihood of gender-neutral words conditioned on gendered sentence prefixes as is shown in Figure FIGREF2 ."}
{"text": "Conclusion and Future Work\tWe demonstrate that a transformer model can be trained to generate questions with correct grammar and relevancy to the context passage and answers provided. WER analyses was applied to diagnose shortcomings and guide future improvements. We observed that a low WER could be due to syntactic similarity but semantic disagreement, while two questions with syntactic divergence but similar meaning could result in a high WER. Since our results does not exhibit issues relating to contextual and syntactic roles of words within a generated question, other popular metrics (BLEU, ROUGE, F1-score, etc.) would lead to similar findings . Perhaps a better approach to evaluating question generation models is to apply state-of-the-art question answering models from SQuAD's leaderboard to measure how many answers agree. To improve the model, more and balanced data can be provided to train the model to reduce the asking word bias. One method that can be used to obtain more data is through data augmentation by back-translation . The original SQuAD can be translated into another language such as French. The translated text could then be translated back into English to generate a variation of the context, question and answers that provide more training data for the model."}
{"text": "Dataset and Features\tWe used dataset, HEOT obtained from one of the past studies done by Mathur et al. where they annotated a set of cleaned tweets obtained from twitter for the conversations happening in Indian subcontinent. A labelled dataset for a corresponding english tweets were also obtained from a study conducted by Davidson et al. This dataset was important to employ Transfer Learning to our task since the number of labeled dataset was very small. Basic summary and examples of the data from the dataset are below:"}
{"text": "Dataset ::: Russian dataset ::: Training Dataset Generation\tWe scraped texts from several pages about civil services from vk.com. Next we extracted frequent $n$-grams that do not contain stopwords and consist of words and digits in various combinations (such as word + 4 digits + word or word + word + 8 digits). We used several rules to merge these $n$-grams so that they resemble real hashtags, for example: remove all whitespace: wordwordworddigits Examples: \u0401\u043b\u043a\u0430\u0412\u0417\u0430\u0437\u0435\u0440\u043a\u0430\u043b\u044c\u0435, \u043d\u0435\u0441\u043a\u043e\u043b\u044c\u043a\u043e\u043b\u0435\u0442\u043d\u0430\u0437\u0430\u0434 replace all whitespace with an underscore: word_word_digits Examples: \u0443\u0432\u0434_\u044e\u0433\u0430_\u0441\u0442\u043e\u043b\u0438\u0446\u044b remove some whitespace and replace other spaces with an underscore: word_worddigits. Examples: \u0438\u0449\u0443\u0441\u0432\u043e\u0435\u0433\u043e\u0433\u0435\u0440\u043e\u044f_\u0443\u0444\u043f\u0441 A word here might be a word in lower case, upper case or capitalized or an abbreviation. There might be up to four digits. In general, we introduced 11 types of hashtags, which contain simply constructed hashtags as well as the complex ones."}
{"text": "$  where $\\delta =1$ for the top two scores and $\\delta =0$ for the rest. $P_k$ is the passage created using both the facts and missing knowledge."}
{"text": "Though, we see that for the proper performance of the therapy is essential a person to support the user to help remember the experiences that are being asked. This project has many possible future lines. In our future work, we suggest to train the model including the Reddit dataset which could improve the chatbot model, as it has many open-domain conversations. Moreover, we would like to include speech recognition and generation, as well as real-time text translation, to make Elisabot more autonomous and open to older adults with reading and writing difficulties. Furthermore, the lack of consistency in the dialogue might be avoided by improving the architecture including information about passed conversation into the model. We also think it would be a good idea to recognize feelings from the user's answers and give a feedback according to them."}
{"text": "In fact, Table TABREF33 shows that the Hack Forums-trained system actually performs better on Nulled, largely due to better performance on previously-seen products. This indicates that there is some truth to Hypothesis 2: product coverage is not the only important factor determining performance."}
{"text": "Overall CoLA Results\tThe overall performance of the three sentence encoders is shown in Table TABREF33 . Performance on CoLA is measured using MCC  . We present the best single restart for each encoder, the mean over restarts for an encoder, and the result of ensembling the restarts for a given encoder, i.e. taking the majority classification for a given sentence, or the majority label of acceptable if tied. For BERT results, we exclude 5 out of the 20 restarts because they were degenerate (MCC=0). Across the board, BERT outperforms GPT, which outperforms the CoLA baseline. However, BERT and GPT are much closer in performance than they are to CoLA baseline. While ensemble performance exceeded the average for BERT and GPT, it did not outperform the best single model."}
{"text": "Qualitative Analysis ::: Limitations\tWe emphasize that QAGS and our overall framework are specifically designed to detect factual inconsistencies in generated summaries relative to the source article. QAGS does not measure other desirable properties of generated text, including fluency, readability, or factual recall. We therefore recommend using QAGS in conjunction with complementary evaluation metrics. The choices of QG and QA models in QAGS are particular to abstractive summarization and may require adaptation to be used for other conditional text generation tasks. For example, we expect that extractive summarization models may obtain nearly perfect QAGS scores because facts and statements are directly copied from the source article."}
{"text": "More generally, an iterated learning process is a repeated zero sum game, in which E makes predictions in virtue of some history, which one might also call a model or a set of hypotheses; if she makes a correct prediction at round n, she reinforces her beliefs in her current history; if she makes a wrong prediction, she adjusts it. The winning condition may be defined in terms of some function of the scores at each learning round or in terms of some global convergence property. Learning conceived in this way is a variant of a simple ME truth game in which costs are assigned to individual discourse moves as in discounted ME games. In an ME truth game, where E develops a history INLINEFORM0 over a set of facts INLINEFORM1 while A argues for an alternative history INLINEFORM2 over INLINEFORM3 , A can successfully defend history INLINEFORM4 as long as either the true history INLINEFORM5 is (a) not learnable or (b) not uniquely learnable. In case (a), E cannot convince the Jury that INLINEFORM6 is the right history; in case (b) A can justify INLINEFORM7 as an alternative interpretation."}
{"text": "Adversarial Attacks in Text\tBecause the purpose of adversarial attacks is to make DNNs misbehave, they can be seen as a classification problem in a broad sense. And majority of recent representative adversarial attacks in text is related to classification so that we categorize them with this feature. In this section, we introduce the majority of existing adversarial attacks in text. Technical details and corresponding comments of each attack method described below are given to make them more clearly to readers."}
{"text": "Results\tThe performances of the NER experiments are reported separately for three different parts of the system proposed. Table 6 presents the comparison of the various methods while performing NER on the bot-generated corpora and the user-generated corpora. Results shown that, in the first case, in the training set the F1 score is always greater than 97%, with a maximum of 99.65%. With both test sets performances decrease, varying between 94-97%. In the case of UGC, comparing the F1 score we can observe how performances significantly decrease. It can be considered a natural consequence of the complex nature of the users' informal language in comparison to the structured message created by the bot. In Table 7, results of the schedule matching are reported. We can observe how the quality of the linking performed by the algorithm is correlated to the choice of the three thresholds. Indeed, the Precision score increase when the time threshold decrease, admitting less candidates as entities during the matching, and when the string similarity thresholds increase, accepting only candidates with an higher degree of similarity. The behaviour of the Recall score is inverted."}
{"text": "We employ a two-stage training procedure, where in the first stage we use $\\vec{D}_{\\text{TTS}}$ to learn the initial parameters as a prior, and then use $\\vec{D}_{\\text{VC}}$ to adapt to the VC model in the second stage. As argued in Section SECREF1, the ability to generate fine-grained hidden representations $\\vec{H}$ is the key to a good VC model, so our goal is to find a set of prior model parameters to train the final encoder $\\text{Enc}^{\\text{S}}_{\\text{VC}}$ and decoder $\\text{Dec}^{\\text{S}}_{\\text{VC}}$. The overall procedure is depicted in Figure FIGREF7."}
{"text": "8)   $C$ is an attention matrix. Note that $\\hat{H^q}$ and $\\hat{H^p}$ is transformed from $H^q$ and $H^p$ by one layer neural network $ReLU(W_3x)$ , respectively. Next, we gather all the information on passages by a simple concatenation of its contextual information $H^p$ and its question-aware representation $H^q \\cdot C$ :  $$U^p = concat(H^p, H^qC) \\in \\mathbb {R}^{4d \\times n}$$   (Eq. 9)  Typically, a passage may contain hundred of tokens, making it hard to learn the long dependencies within it. Inspired by  , we apply a self-attended layer to rearrange the information $U^p$ as:  $$\\hat{U}^p = U^p drop_{diag}(f_{attention}(U^p, U^p)).$$   (Eq."}
{"text": "Predicting political views\tParties change their policies and positions in the political spectrum. More reliable categories for political bias are party independent labels for political views, see sec:data. A separate suite of experiments was run to train and test the prediction performance of the text classifiers models described in sec:model. As there was no out-of-domain evaluation set available in this setting only evaluation error on in-domain data is reported. Note however that also in this experiment the evaluation data was never seen by any model during training time. In tab:resultsavgpoliticalview results for the best and worst classes, in terms of predictability, are listed along with the average performance metrics on all classes. Precision and recall values of close to 0.5 on average can be considered rather high considering the large number of labels."}
{"text": "Joint-BiLSTM: Different from S2VT-based models, we employ a joint bidirectional LSTM networks to encode video sequence and decode description applying another LSTM respectively rather than sharing the common one. We stack two layers of LSTM networks to encode video and pre-train language model as in S2VT-based models. Similarly, unidirectional LSTM, bidirectional LSTM and reinforced BiLSTM are executed to investigate the performance of each structure. We set 1024 hidden units of the first LSTM in unidirectional encoder so that the output could pass to the second encoder directly, and the memory cell and hidden state of the last time point are applied to initialize description decoder. Bidirectional structure and reinforced BiLSTM in encoder are implemented similarly to the corresponding type structure in S2VT-based models, respectively, and then feed the video representation into description generator as the unidirectional model aforementioned."}
{"text": "Small-Footprint Keyword Spotting on Raw Audio Data with Sinc-Convolutions\tKeyword Spotting (KWS) enables speech-based user interaction on smart devices. Always-on and battery-powered application scenarios for smart devices put constraints on hardware resources and power consumption, while also demanding high accuracy as well as real-time capability. Previous architectures first extracted acoustic features and then applied a neural network to classify keyword probabilities, optimizing towards memory footprint and execution time. Compared to previous publications, we took additional steps to reduce power and memory consumption without reducing classification accuracy. Power-consuming audio preprocessing and data transfer steps are eliminated by directly classifying from raw audio. For this, our end-to-end architecture extracts spectral features using parametrized Sinc-convolutions. Its memory footprint is further reduced by grouping depthwise separable convolutions. Our network achieves the competitive accuracy of 96.4% on Google's Speech Commands test set with only 62k parameters."}
{"text": "You can notice that we increased the number of languages from the first review of our paper. In addition, the table shows the comparison between our baseline method (Least Squares) and our final one (Normalized vectors, with cosine similarity), which shows a slight improvement with respect to the first one."}
{"text": "Discussion & Conclusions\tWe established that our Siamese recurrent networks (with SRN, GRU or LSTM cells) are able to recognize logical entailment relations without any a priori cues about syntax or semantics of the input expressions. Indeed, some of the recurrent set-ups even outperform tree-shaped networks, whose topology is specifically designed to deal with such tasks. This indicates that recurrent networks can develop representations that can adequately process a formal language with a nontrivial hierarchical structure. The formal language we defined did not exploit the full expressive power of first-order predicate logic; nevertheless by using standard first-order predicate logic, a standard theorem prover, and a set-up where the training set only covers a tiny fraction of the space of possible logical expressions, our experiments avoid the problems observed in earlier attempts to demonstrate logical reasoning in recurrent networks. The experiments performed in the last few sections moreover show that the GRU and LSTM architectures exhibit at least basic forms of compositional generalization. In particular, the results of the zero-shot generalization experiments with novel lengths and novel words cannot be explained with a `memorize-and-interpolate' account, i.e."}
{"text": "However when performing the same comparison between our large models (see Table TABREF24), this performance gap becomes significantly smaller. As expected, not using the multi-span head causes the multi-span performance to plummet. Note that for this ablation test the single-span heads were permitted to train on multi-span questions. Compared to using greedy decoding in the prediction of multi-span questions, using beam search results in a small improvement. We used a beam with of 5, and didn't perform extensive tuning of the beam width."}
{"text": "Corpus Annotation ::: Extraction of PIE Candidates\tTo annotate the corpus completely manually would require annotators to read the whole corpus, and cross-reference each sentence to a list of almost 600 PIEs, to check whether one of those PIEs occurs in a sentence. We do not consider this a feasible annotation settings, due to both the difficulty of recognising literal usages of idioms and the time cost needed to find enough PIEs, given their low overall frequency. As such, we use a pre-extraction step to present candidates for annotation to the human annotators. Given the corpus and the set of PIEs, we heuristically extract the PIE candidates as follows: given an idiomatic expression, extract every sentence which contains all the defining words of the idiom, in any form. This ensures that all possibly matching sentences get extracted, while greatly pruning the amount of sentences for annotators to look at. In addition, it allows us to present the heuristically matched PIE type and corresponding words to the annotators, which makes it much easier to judge whether something is a PIE or not. This also means that annotators never have to go through the full list of PIEs during the annotation process."}
{"text": "Experiments and Results ::: Experimental Settings\tData Splits: We used the same splits of train, develop, and test sets used by Karlekar and Bansal , with 7201, 990 and 1701 stories, respectively. In this study, we only considered single label classifications. Baseline Models: CNN and BiLSTM models that perform classification and extraction separately were used as baseline models. In classification, we also experimented with BiLSTM with the attention layer. To demonstrate that the improvement came from joint learning structure rather the two layer structure in J-CNN, we investigated the same model structure without training on key element extraction. We use J-CNN* to denote it. Preprocess: All the texts were converted to lowercase and preprocessed by removing non-alphanumeric characters, excluding \u201c. ! ? \u201d . The word embeddings were pre-trained using fastText  with dimension equaling 100. Hyperparameters: For the CNN model, the filter size was chosen to be (1,2,3,4), with 50 filters per filter size. Batch size was set to 50 and the dropout rate was 0.5."}
{"text": "Proposed System ::: The Crawler ::: Decide\tA decider has three main decisions to take. First, based on the metadata associated with an URL, should it be visited? In practice, we visit only new URLs, but the tool is designed in a way such that a recrawl is possible if the page is detected as highly dynamic. The second decision arises at the end of the processing, where the page can be either saved or blacklisted. To favor recall, we currently keep any URL with at least one GSW sentence. Finally, the decider can choose to visit the outgoing links or not. After some trials, we found that following links from pages with more than two new GSW sentences is a reasonable choice, as pages with less sentences are often quotes or false positives."}
{"text": "It is the same with the second example, where a human selects \u201cUTF8gbsn\u4e13\u5bb6\u201d (expert) and Dual-Train selects \u201cUTF8gbsn\u5de5\u4f5c\u8005\u201d (worker), while the baseline selects \u201cUTF8gbsn\u94bb\u7814\u201d (research) and fails to generate a grammatical sentence later. In the third one, the reference and the baseline use the same word, while Dual-Train chooses a word of the same meaning. It can be concluded that Dual-Train indeed learns better word relations that could generalize to the test set, and good word relations can guide the decoder to generate semantically consistent summaries. To show why the generated text of the proposed method is more related to the source content, we further analyze the label distribution, i.e., the word distribution, generated by the (first) output layer, from which the output word is selected. To illustrate the relationship, we calculate a representation for each word based on the label distributions. Each representation is associated with a specific label (word), denoted by INLINEFORM0 , and each dimension INLINEFORM1 shows how likely the label indexed by INLINEFORM2 will be generated instead of the label INLINEFORM3 ."}
{"text": "We mask (exclude) all words of the vocabulary except the current source sentence tokens, relation tokens, separator tokens (`;', `$\\vert $'), UNK, and EOS tokens in the softmax operation. To mask (exclude) some word from softmax, we set the corresponding value in $\\hat{\\mathbf {o}}_t$ at $-\\infty $ and the corresponding softmax score will be zero. This ensures the copying of entities from the source sentence only. We include the UNK token in the softmax operation to make sure that the model generates new entities during inference. If the decoder predicts an UNK token, we replace it with the corresponding source word which has the highest attention score. During inference, after decoding is finished, we extract all tuples based on the special tokens, remove duplicate tuples and tuples in which both entities are the same or tuples where the relation token is not from the relation set. This model is referred to as WordDecoding (WDec) henceforth."}
{"text": "We denote this categorical variable as the \"head predictor\". Numerically-aware BERT (NABERT+)  introduced two main improvements over NAQANET. The first was to replace the QANET encoder with BERT. This change alone resulted in an absolute improvement of more than eight points in both EM and F1 metrics. The second improvement was to the arithmetic head, consisting of the addition of \"standard numbers\" and \"templates\". Standard numbers were predefined numbers which were added as additional inputs to the arithmetic head, regardless of their occurrence in the passage. Templates were an attempt to enrich the head's arithmetic capabilities, by adding the ability of doing simple multiplications and divisions between up to three numbers. MTMSN  is the first, and only model so far, that specifically tried to tackle the multi-span questions of DROP. Their approach consisted of two parts. The first was to train a dedicated categorical variable to predict the number of spans to extract. The second was to generalize the single-span head method of extracting a span, by utilizing the non-maximum suppression (NMS) algorithm  to find the most probable set of non-overlapping spans."}
{"text": "Acknowledgments\tThis research was supported by the Swiss National Science Foundation (SNSF) grant number 407540_167176 under the project \"Conversational Agent for Interactive Access to Information\"."}
{"text": "In this way, we filter out i) articles with excessively long content to reduce memory consumption; ii) very short leading sentences with little information which are unlikely to be a good summary. To encourage the model to generate abstrative summaries, we also remove articles where any of the top three sentences is exactly repeated in the rest of the article. Third, we try to remove articles whose top three sentences may not form a relevant summary. For this purpose, we utilize a simple metric: overlapping words. We compute the portion of non-stopping words in the top three sentences that are also in the rest of an article. A higher portion implies that the summary is representative and has a higher chance of being inferred by the model using the rest of the article. To verify, we compute the overlapping ratio of non-stopping words between human-edited summary and the article in CNN/DailyMail dataset, which has a median value of 0.87. Therefore, in pretraining, we keep articles with an overlapping word ratio higher than 0.65."}
{"text": "Modelos propuestos ::: Modelo generativo estoc\u00e1stico usando cadenas de Markov\tEste modelo generativo, que llamaremos Modelo de Markov, est\u00e1 basado en el algoritmo de Viterbi y las cadenas de Markov , donde se selecciona una etiqueta POS con la m\u00e1xima probabilidad de ocurrencia, para ser agregada al final de la secuencia actual. Utilizamos el corpus de frases literarias 8KF (ver Secci\u00f3n SECREF5), que fue convenientemente filtrado para eliminar tokens indeseables: n\u00fameros, siglas, horas y fechas. El corpus filtrado se analiz\u00f3 usando Freeling, que recibe en entrada una cadena de texto y entrega el texto con una etiqueta POS para cada palabra. El corpus es analizado frase a frase, reemplazando cada palabra por su respectiva etiqueta POS. Al final del an\u00e1lisis, se obtiene un nuevo corpus 8KPOS con $s = 7~679$ secuencias de etiquetas POS, correspondientes al mismo n\u00famero de frases del corpus 8KF."}
{"text": "Our classifier features: In the following sections, we discuss related work on hate speech classification, followed by a description of the datasets, methods and results of our study."}
{"text": "Experiments ::: Baselines ::: CVAE.\tAn RNN-based conditional variational autoencoder for dialogue response generation , which uses a multivariate Gaussian latent variable to model the response and concatenate it with the last hidden state of the encoder as the initial state of the decoder. KL annealing, early stopping strategy and bag-of-word auxiliary loss are applied during the training. We use the implementation released by ."}
{"text": "Type Classification\tWe employ a multi-class Naive Bayes classifier as the second stage classification mechanism, for categorizing tweets appropriately, depending on the type of emergencies they indicate. This multi-class classifier is trained on data manually labeled with classes. We tokenize the training data using \u201cNgramTokenizer\u201d and then, apply a filter to create word vectors of strings before training. We use \u201ctrigrams\u201d as features to build a model which, later, classifies tweets into appropriate categories, in real time. We then perform cross validation using standard techniques to calculate the results, which are shown under the label \u201cStage 2\u201d, in table TABREF20 ."}
{"text": "Methodology ::: Interval Bound Propagation ::: Verifiable Training.\tThe upper bound in (DISPLAY_FORM17) is fast to compute (only requires two forward passes for upper and lower bounds through the network). Hence, we can define a loss to optimise models such that the models are trained to be verifiable. Solving (DISPLAY_FORM17) is equivalent to finding the worst-case logit difference, and this is achieved when the logit of the true class is equal to its lower bound, and all other logits equal to their upper bounds. Concretely, for each class $y \\ne y_\\textrm {true} $: $\\hat{\\mathbf {z}}_{K,y}(\\delta ) = \\overline{\\mathbf {z}}_{K,y} (\\delta ) $, and $\\hat{\\mathbf {z}}_{K,y_\\textrm {true}}(\\delta ) = \\underline{\\mathbf {z}}_{K,y_\\textrm {true}} (\\delta ) $."}
{"text": "Interesting directions of future work include: tightening IBP bounds to allow applicability to deeper models, investigating bound propagation in other types of neural architectures (e.g. those based on recurrent networks or self-attention), and exploring other forms of specifications in NLP."}
{"text": "Methods ::: Incorporating Commonsense Rationales\tWe explore how to utilize additional commonsense knowledge (i.e. rationales) as the input to the task. Like we mentioned in Section SECREF6, we search relevant sentences from the OMCS corpus as the additional distant rationales, and ground truth rationale sentences for dev/test data. The inputs are no longer the concept-sets themselves, but in a form of \u201c[rationales$|$concept-set]\u201d (i.e. concatenating the rationale sentences and original concept-set strings)."}
{"text": "Queries, on the other hand, are not expressed in natural language, but instead consist of tuples $q$2 where the object entity is unknown and it has to be inferred by reading the support documents. Therefore, answering a query corresponds to finding the entity $q$3 that is the object of a tuple in the KB with subject $q$4 and relation $q$5 among the provided set of candidate answers $q$6 . The goal is to learn a model that can identify the correct answer $a^\\star $ from the set of supporting documents $S_q$ . To that end, we exploit the available supervision to train a neural network that computes scores for candidates in $C_q$ . We estimate the parameters of the architecture by maximizing the likelihood of observations. For prediction, we then output the candidate that achieves the highest probability. In the following, we present our model discussing the design decisions that enable multi-step reasoning and an efficient computation."}
{"text": "Approach and Contributions\tOur approach to generating question-answer pairs from text is a two-stage process: in the first stage we select the most relevant and appropriate candidate answer, i.e., the pivotal answer, using an answer selection module, and in the second stage we encode the answer span in the sentence and use a sequence to sequence model with a rich set of linguistic features to generate questions for the pivotal answer. Our sentence encoder transforms the input sentence into a list of fixed-length continuous vector word representation, each input symbol being represented as a vector. The question decoder takes in the output from the sentence encoder and produces one symbol at a time and stops at the EOS (end of sentence) marker. To focus on certain important words while generating questions (decoding) we use a global attention mechanism. The attention module is connected to both the sentence encoder as well as the question decoder, thus allowing the question decoder to focus on appropriate segments of the sentence while generating the next word of the question. We include linguistic features for words so that the model can learn more generalized syntactic transformations. We provide a detailed description of these modules in the following sections."}
{"text": "Conclusions\tWe introduced the problem of parsing AMR structures, annotated for English, from sentences written in other languages as a way to test the cross-lingual properties of AMR. We provided evidence that AMR can be indeed shared across the languages tested and that it is possible to overcome translational divergences. We further proposed a novel way to evaluate the target parsers that does not require manual annotations of the target language. The full-cycle procedure is not limited to AMR parsing and could be used for other cross-lingual problems in NLP. The results of the projection-based AMR parsers indicate that there is a vast room for improvements, especially in terms of generating better alignments. We encourage further work in this direction by releasing professional translations of the AMR test set into four languages."}
{"text": "Approximate inference\tGiven a dataset INLINEFORM0 , the goal of inference is to compute the posterior distribution of the per-document topic proportions INLINEFORM1 , the per-word topic assignments INLINEFORM2 , the per-topic distribution over words INLINEFORM3 , the per-document latent true class INLINEFORM4 , and the per-annotator confusion parameters INLINEFORM5 . As with LDA, computing the exact posterior distribution of the latent variables is computationally intractable. Hence, we employ mean-field variational inference to perform approximate Bayesian inference. Variational inference methods seek to minimize the KL divergence between the variational and the true posterior distribution. We assume a fully-factorized (mean-field) variational distribution of the form DISPLAYFORM0   where INLINEFORM0 , INLINEFORM1 , INLINEFORM2 , INLINEFORM3 and INLINEFORM4 are variational parameters. Table TABREF23 shows the correspondence between variational parameters and the original parameters. Let INLINEFORM0 denote the model parameters. Following  , the KL minimization can be equivalently formulated as maximizing the following lower bound on the log marginal likelihood DISPLAYFORM0   which we maximize using coordinate ascent. Optimizing INLINEFORM0 w.r.t."}
{"text": "We obtain samples from $\\mathbf {z}|\\mathbf {x},\\mathbf {y}$ by performing collapsed Gibbs sampling as described in section 5.1 of journals/jasa/Teh06: we repeatedly sample $\\mathbf {z}$ from its conditional distribution, with $\\mathbf {\\phi }$ integrated out:  $$z^\\textbf {n}_i | \\mathbf {x}, \\mathbf {y}, z^\\textbf {n}_{-i} ="}
{"text": "Introduction\tThe collection and analysis of historical document images is a key component in the preservation of culture and heritage. Given its importance, a number of active research efforts exist across the world , , , , , . In this paper, we focus on palm-leaf and early paper documents from the Indian sub-continent. In contrast with modern or recent era documents, such manuscripts are considerably more fragile, prone to degradation from elements of nature and tend to have a short shelf life , , . More worryingly, the domain experts who can decipher such content are small in number and dwindling. Therefore, it is essential to access the content within these documents before it is lost forever. Surprisingly, no large-scale annotated Indic manuscript image datasets exist for the benefit of researchers in the community. In this paper, we take a significant step to address this gap by creating such a dataset. Given the large diversity in language, script and non-textual regional elements in these manuscripts, spatial layout parsing is crucial in enabling downstream applications such as OCR, word-spotting, style-and-content based retrieval and clustering. For this reason, we first tackle the problem of creating a diverse, annotated spatial layout dataset."}
{"text": "Users open some existing documents in a file system, make some changes (e.g. addition, deletion or replacement), and save them as separate documents. These documents become revisions of the original documents. We started from an initial corpus that did not have revisions, and kept adding new documents and revising existing documents. Similar to a file system, at any moment new documents could be added and/or some of the current documents could be revised. The revision operations we used were deletion, addition and replacement of words, sentences, paragraphs, section names and document titles. The addition of words, ..., section names, and new documents were pulled from the Wikipedia abstracts. This corpus generation process had five time periods INLINEFORM0 . Figure FIGREF42 illustrates this simulation. We set a Poisson distribution with rate INLINEFORM1 (the number of documents in the initial corpus) to control the number of new documents added in each time period, and a Poisson distribution with rate INLINEFORM2 to control the number of documents revised in each time period. We generated six data sets using different random seeds, and each data set contained six corpora (Corpus 0 - 5)."}
{"text": "As we train with more data, we obtain better validation losses. This was expected. The second trend is that by using less than 50% of the data available the model tends to overfit the data, as indicated by the consistent increase in the validation loss after about 15 epochs (check dashed lines in right side of Figure FIGREF28 ). This suggests that for the future we should not try any drastic reduction of the training data to save training time. Finally, when not overfitting, the validation loss seems to stabilize after around 20 epochs. We observed no phase-transition effects (the model seems simple enough for not showing that type of behavior). This indicates we have a practical way of safely deciding when to stop training the model."}
{"text": "Introduction\tCurrently, social networks are so popular. Some of the biggest ones include Facebook, Twitter, Youtube,... with extremely number of users. Thus, controlling content of those platforms is essential. For years, social media companies such as Twitter, Facebook, and YouTube have been investing hundreds of millions euros on this task , . However, their effort is not enough since such efforts are primarily based on manual moderation to identify and delete offensive materials. The process is labour intensive, time consuming, and not sustainable or scalable in reality , , . In the sixth international workshop on Vietnamese Language and Speech Processing (VLSP 2019), the Hate Speech Detection (HSD) task is proposed as one of the shared-tasks to handle the problem related to controlling content in SNSs. HSD is required to build a multi-class classification model that is capable of classifying an item to one of 3 classes (hate, offensive, clean)."}
{"text": "We suspect that the less drastic the change in an attention dimension over time, the higher the chances of that dimension being part of multiple cross-view dynamics. Thus more attentions activate this important dimension. O4: Some attentions focus on cross-view dynamics that involve only two modalities. For example, in $a^3$ , the audio modality has no dark blue dimensions, while in $a^1$ all the modalities have dark blue dimensions. The attentions seem to have residual effects. $a^1$ shows activations over a broad set of variables while $a^4$ shows activation for fewer sets, indicating that attentions could learn to act in a complementary way."}
{"text": "Reinforcement Learning\tIn the last few years, reinforcement learning has emerged as a powerful tool for solving complex sequential decision-making problems. It is well known for its great success in the game field, such as Go  and Atari games  . Recently, reinforcement learning has also been successfully applied to many natural language processing tasks and achieved good performance  ,  ,  . Feng et al.  used reinforcement learning for relation classification task by filtering out the noisy data from the sentence bag and they achieved huge improvements compared with traditional classifiers. Zhang et al.  applied the reinforcement learning on sentence representation by automatically discovering task-relevant structures. To automatic taxonomy induction from a set of terms, Han et al.  designed an end-to-end reinforcement learning model to determine which term to select and where to place it on the taxonomy, which effectively reduced the error propagation between two phases. Inspired by the above works, we also add reinforcement learning to our framework."}
{"text": "In multitask learning, a set of related tasks are learned (e.g., emotional activation), along with a primary task (e.g., emotional valence); both tasks share parts of the network topology and are hence jointly trained, as depicted in Figure FIGREF4 . It is expected that data for the secondary task models information, which would also be discriminative in learning the primary task. In fact, this approach has been shown to improve generalizability across corpora  . The remainder of this paper is organized as follows: First we introduce the DCGAN model and discuss prior work, in Section SECREF2 . Then we describe our specific multitask DCGAN model in Section SECREF3 , introduce the datasets in Section SECREF4 , and describe our experimental design in Section SECREF5 . Finally, we report our results in Section SECREF6 and discuss our findings in Section SECREF7 ."}
{"text": "Introduction\tIn recent years, word embeddings , ,  have been proven to be very useful for training downstream natural language processing (NLP) tasks. Moreover, contextualized embeddings ,  have been shown to further improve the performance of NLP tasks such as named entity recognition, question answering, or text classification when used as word features because they are able to resolve ambiguities of word representations when they appear in different contexts. Different deep learning architectures such as multilingual BERT , LASER  and XLM  have proved successful in the multilingual setting. All these architectures learn the semantic representations from unannotated text, making them cheap given the availability of texts in online multilingual resources such as Wikipedia. However, the evaluation of such resources is usually done for the high-resourced languages, where one has a smorgasbord of tasks and test sets to evaluate on. This is the best-case scenario, languages with tones of data for training that generate high-quality models. For low-resourced languages, the evaluation is more difficult and therefore normally ignored simply because of the lack of resources."}
{"text": "Performing curriculum learning strategy offers a considerable improvement, take the topmost generating hierarchy in Table TABREF13 for example, this method yields a 102.07% improvement in BLEU, 48.26% in ROUGE-1, 144.8% in ROUGE-2, and 39.18% in ROUGE-L. Despite that applying repeat-input mechanism alone does not offer benefit, combining these two strategies together further achieves the best performance. Note that these methods do not require any additional parameters. Unfortunately, even some of the attentional hierarchical decoders achieve the best results in the generating hierarchies (Table TABREF18 ). Mostly, the additional attention mechanisms are not capable of bringing benefit for model performance. The reason may be that the decoding process is designed for gradually importing words in the specific set of linguistic patterns to the output sequence, each decoder layer is responsible of copying the output tokens from the previous layer and insert new words into the sequence precisely. Because of this nature, a decoder needs explicit information of the structure of a sentence rather than implicit high-level latent information."}
{"text": "Nonetheless, few attempts have been made to distinguish both categories from another perspective. Potha and Stamatatos  , for example, categorize AV methods as either intrinsic or extrinsic (emphasis by us): \u201cIntrinsic verification models view it [i. e., the verification task] as a one-class classification task and are based exclusively on analysing the similarity between [ INLINEFORM0 ] and [ INLINEFORM1 ]. [...] Such methods [...] do not require any external resources.\u201d   \u201cOn the other hand, extrinsic verification models attempt to transform the verification task to a pair classification task by considering external documents to be used as samples of the negative class.\u201d   While we agree with statement (2), the former statement (1) is unsatisfactory, as intrinsic verification models are not necessarily unary. For example, the AV approach GLAD proposed by H\u00fcrlimann et al.  directly contradicts statement (1). Here, the authors \u201cdecided to cast the problem as a binary classification task where class values are Y [ INLINEFORM0 ] and N [ INLINEFORM1 ]. [...] We do not introduce any negative examples by means of external documents, thus adhering to an intrinsic approach.\u201d  ."}
{"text": "This variation is due to the different ways in which annotators were recruited: student-assistants were recruited for several days, while participants recruited through the Radboud Research Participation System could only join for the duration of an hour. We calculated inter-annotator agreement by Krippendorff's Alpha , which accounts for different annotator pairs and empty values. To also zoom in on the particular agreement by category, we calculated mutual F-scores for each of the categories. This metric is typically used to evaluate system performance by category on gold standard data, but could also be applied to annotation pairs by alternating the roles of the two annotators between classifier and ground truth. A summary of the agreement by categorization is given in Table TABREF10. While both the Relevance and Subject categorizations are annotated at a percent agreement of $0.71$ and $0.70$, their agreement scores are only fair, at $\\alpha =0.27$ and $\\alpha =0.29$. The percent agreement on Stance and Sentiment, which carry more categories than the former two, is $0.54$ for both."}
{"text": "Results and Discussion ::: Performance analysis\tIn Figure FIGREF14, we show some of the improvements in performance obtained when including a fixed amount of virtual edges using GloVe as embedding method. In each subpanel, we show the relative improvement in performance obtained as a function of the fraction of additional edges. In this section, we considered the traditional co-occurrence as starting point. In other words, the network construction disregarded stopwords. The list of stopwords considered in this paper is available in the Supplementary Information. We also considered the global approach to filter edges. The relative improvement in performance is given by $\\Gamma _+{(p)}/\\Gamma _0$, where $\\Gamma _+{(p)}$ is the accuracy rate obtained when $p\\%$ additional edges are included and $\\Gamma _0 = \\Gamma _+{(p=0)}$, i.e. $\\Gamma _0$ is the accuracy rate measured from the traditional co-occurrence model. We only show the highest relative improvements in performance for each classifier. In our analysis, we considered also samples of text with distinct length, since the performance of network-based methods is sensitive to text length ."}
{"text": "Proposed Framework ::: Captioning Module\tIn this section we explain the captioning based for an example modality, namely, visual. Given a video $V$ and a set of proposals $P_V$ from the event localization module, the task of the captioning module is to provide a caption for each proposal in $P_V$. In order to extract features from a video $V$, we employ I3D network  pre-trained on the Kinetics dataset which produces 1024-d features. The gap between the extracted features and the generated captions is filled with Transformer  architecture which was proven to effectively encode and decode the information in a sequence-to-sequence setting."}
{"text": "In practice, given the size of our dataset, calculating the full denominator of the loss for a query, image pair would involve iterating through each image for each query, which is $O(n^2)$ in the number of training examples. To remedy this, we calculated the loss within each batch separately. That is, the denominator of the loss only involved summing over images in the same batch as the query. We used a batch size of 1000 for all experiments. In principle, the negative sampling approach used by  could be used instead to prevent quadratic time complexity. We can interpret this loss function as producing a softmax classification task for queries and images: given a query, the model needs to predict the image relevant to that query. The cosine similarity between the image representation $I_i$ and the query representation $Q_q$ is normalized under softmax to produce a \u201cbelief\u201d that the image $i$ is the image relevant to the query $q$ . This is analogous to the skip-gram model proposed by  , although we use cosine similarity instead of dot product."}
{"text": "The problem could arise from the fact that words as \u201cUTF8gbsn\u671f\u8d27\u4e1a\u201d rarely occur in the training data, and their relatedness is not reflected in the data. Another issue is that there are some particles, e.g., \u201cUTF8gbsn\u7684\u201d (DE) in the most related words. A possible explanation is that particles show up too often in the contexts of the word, and it is hard for the models to distinguish them from the real semantically-related words. As our proposed approach is based on regularization of the less common correspondence, it is reasonable that such kind of relation cannot be eliminated. The first case can be categorized into data sparsity, which usually needs the aid of knowledge bases to solve. The second case is due to the characteristics of natural language. However, as such words are often closed class words, the case can be resolved by manually restricting the relatedness of these words."}
{"text": "Experiments ::: Model Analysis ::: Examples with Different Learning Frequencies\tAs shown in Table TABREF30, the most frequently learnt examples are comprehensively far better than those seldom learnt examples, which exhibits the effectiveness of the adaptive multi-curricula learning framework."}
{"text": "Acknowledgments\tWe would like to thank the anonymous reviewers for their comments to improve this paper. We would also like to thank Sebastian Pad\u00f3 for his help training the Clark clusters. This work has been supported by the European projects NewsReader, EC/FP7/316404 and QTLeap - EC/FP7/610516, and by the Spanish Ministry for Science and Innovation (MICINN) SKATER, Grant No. TIN2012-38584-C06-01 and TUNER, TIN2015-65308-C5-1-R."}
{"text": "Analysis of training datasets ::: Annotation of training datasets ::: Identity of the annotators\tThe data statements framework given by Bender and Friedman emphasises the importance of understanding who has completed annotations. Knowing who the annotators are is important because `their own \u201csocial address\" influences their experience with language and thus their perception of what they are annotating.'  In the context of online abuse, Binns et al. show that the gender of annotators systematically influences what annotations they provide . No annotator will be well-versed in all of the slang or coded meanings used to construct abusive language. Indeed, many of these coded meanings are deliberately covert and obfuscated . To help mitigate these challenges, annotators should be (a) well-qualified and (b) diverse. A homogeneous group of annotators will be poorly equipped to catch all instances of abuse in a corpus. Recruiting an intentionally mixed groups of annotators is likely to yield better recall of abuse and thus a more precise dataset . Information about annotators is unfortunately scarce."}
{"text": "Analogously to the work of Pelevina:16, we construct a semantic relatedness graph from a list of nearest neighbours, but we filter this list using the following procedure: Extract a list $\\mathcal {N}$ = {$w_{1}$, $w_{2}$, ..., $w_{N}$} of $N$ nearest neighbours for the target (ego) word vector $w$. Compute a list $\\Delta $ = {$\\delta _{1}$, $\\delta _{2}$, ..., $\\delta _{N}$} for each $w_{i}$ in $\\mathcal {N}$, where $\\delta _{i}~=~w-w_{i}$. The vectors in $\\delta $ contain the components of sense of $w$ which are not related to the corresponding nearest neighbours from $\\mathcal {N}$. Compute a list $\\overline{\\mathcal {N}}$ = {$\\overline{w_{1}}$, $\\overline{w_{2}}$, ..., $\\overline{w_{N}}$}, such that $\\overline{w_{i}}$ is in the top nearest neighbours of $\\delta _{i}$ in the embedding space."}
{"text": "Challenge Details\tA dedicated website for the competition was set up. The website included instructions, the registration form, schedule, and other relevant details. Following registration, participants were able to download the training datasets. The label distribution of emotions in our data are highly unbalanced, as can be seen in Figure FIGREF6. Due to the small number of three of the labels, participants were instructed to use only four emotions for labels: joy, sadness, anger, and neutral. Evaluation of submissions was done using only utterances with these four labels. Utterances with labels other than the above four (i.e., surprise, disgust, fear or non-neutral) were discarded and not used in the evaluation. Scripts for verifying and evaluating the submissions were made available online. We used micro-F1 as the comparison metric."}
{"text": "The quality of our model\u2014the conversion tool\u2014comes from how well we encode prior knowledge about the relationship between the UD and UniMorph corpora."}
{"text": "MT performed the best on the test examples which according to the annotation require a correct understanding of the vocabulary (81 $\\text{F}_1$ on binary /54 $\\text{F}_1$ on 4-class), with Blse (79/48) slightly worse. Muse (76/23), VecMap (70/35), and Barista (67/41) perform worse. This suggests that Blse is better than Muse, VecMap and Barista at transferring sentiment of the most important sentiment bearing words. Negation: Negation is a well-studied phenomenon in sentiment analysis Pang2002,Wiegand2010,Zhu2014,Reitan2015 . Therefore, we are interested in how these four models perform on phrases that include the negation of a key element, for example \u201cIn general, this hotel isn't bad\". We would like our models to recognize that the combination of two negative elements \u201cisn't\" and \u201cbad\" lead to a Positive label. Given the simple classification strategy, all models perform relatively well on phrases with negation (all reach nearly 60 $\\text{F}_1$ in the binary setting)."}
{"text": "Related Work ::: Training on synthetic data\tThe lack of training data is an issue for many NLP applications. There have been attempts to generate and use synthetic data for training question answering systems  and SQL2text systems . In  synthetic hashtags are generated by removing whitespace characters from frequent n-grams, while in  German compounds are synthesized for further machine translation."}
{"text": "Introduction\tMachine translation (MT) research is biased towards language pairs including English due to the ease of collecting parallel corpora. Translation between non-English languages, e.g., French$\\rightarrow $German, is usually done with pivoting through English, i.e., translating French (source) input to English (pivot) first with a French$\\rightarrow $English model which is later translated to German (target) with a English$\\rightarrow $German model , , . However, pivoting requires doubled decoding time and the translation errors are propagated or expanded via the two-step process. Therefore, it is more beneficial to build a single source$\\rightarrow $target model directly for both efficiency and adequacy. Since non-English language pairs often have little or no parallel text, common choices to avoid pivoting in NMT are generating pivot-based synthetic data ,  or training multilingual systems , . In this work, we present novel transfer learning techniques to effectively train a single, direct NMT model for a non-English language pair. We pre-train NMT models for source$\\rightarrow $pivot and pivot$\\rightarrow $target, which are transferred to a source$\\rightarrow $target model."}
{"text": "Such a model concatenates the hidden state from LSTM and a vector constructed by self-attention at each position, and feeds them into CRF as features. The model attempts to capture rich implicit structures in the input space, but it does not put effort on explicit structures in the output space. E-I is a weaker version of EI. Such a model removes the BMES sub-tags in the E tag, causing the model to learn less explicit structural information in the output space. EI- is a weaker version of EI. Such a model removes the self-attention from EI, causing the model to learn less expressive implicit structures in the input space."}
{"text": "Conclusion\tNamed entities being an open set which keeps expanding are difficult to represent through a closed NE dictionary. This work mitigates significant defects in previous closed NE definitions and proposes a new open definition for NEs by modeling their embedding distributions with least parameters. We visualize NE distributions in monolingual case and perform an effective isomorphism spaces mapping in cross-lingual case. According to our work, we demonstrate that common named entity types (PER, LOC, ORG) tend to be densely distributed in a hypersphere and it is possible to build a mapping between the NE distributions in embedding spaces to help cross-lingual NE recognition. Experimental results show that the distribution of named entities via mapping can be used as a good enough replacement for the original distribution. Then the discovery is used to build an NE dictionary for Indonesian being a truly low-resource language, which also gives satisfactory precision."}
{"text": "Evaluation of Representation Models ::: Results ::: The unsupervised scenario.\tTable TABREF26 reports the results using the item representations directly. We report results over train and dev partitions for completeness, but note that none of them was used to tune the models. As it can be seen, multimodal representations consistently outperform their text-only counterparts. This confirms that, overall, visual information is helpful in the semantic textual similarity task and that image and sentence representation are complementary. For example, the bert model improves more than 13 points when visual information provided by the resnet is concatenated. glove shows a similar or even larger improvement, with similar trends for use and vse++(text). Although vse++(img) shows better performance than resnet when applying them alone, further experimentation showed lower complementarity when combining with textual representation (e.g. $0.807\\rho $ in test combining textual and visual modalities of vse++). This is something expected as vse++(img) is pre-trained along with the textual part of the vse++ model on the same task. We do not show the combinations with vse++(img) due to the lack of space."}
{"text": "Analogy Datasets\tTable 3 compares Chinese word analogy datasets. Most analogies in existing datasets are morphological (morph.) or named entity (entity) relations. For example, CA8-Morphological  uses 21 shared prefix characters, e.g., UTF8bkai\u7b2c, to form 2,553 analogies, e.g., UTF8bkai\u4e00 : UTF8bkai\u7b2c\u4e00 = UTF8bkai\u4e8c : UTF8bkai\u7b2c\u4e8c (one : first = two : second). As for named entities, some 20 word pairs of the capital-country relation can be permuted to form 190 analogies, which require a knowledge base but not commonsense to solve. Only the nature part of CA8 and the man-woman part of CA-Google  contains a handful of relations that requires commonsense world knowledge. In constrast, CA-EHN extracts 85K linguist-checked analogies covering 6,490 concept pairs, e.g., (wood $\\vert $ UTF8bkai\u6728, UTF8bkai\u99ac $\\vert $ horse)."}
{"text": "The strongest coalitions on the left-to-center part of the axis are those between GUE-NGL, Greens-EFA, and S&D, and between S&D, ALDE, and EPP. These results reaffirm the role of Greens-EFA and ALDE as intermediaries, not only in the European Parliament but also in the debates on social media. Last, but not least, with the ERGM methodology we measure the extent to which the retweet network can explain the co-voting activities in the European Parliament. We compute this for each policy area separately and also over all RCVs. We conclude that the retweet network indeed matches the co-voting behavior, with the exception of one specific policy area. In the area Economic and monetary system, the links in the (overall) retweet network do not match the links in the co-voting network. Moreover, the negative coefficients imply a radically different formation of coalitions in the European Parliament. This is consistent with the results in Figs FIGREF36 and FIGREF37 (the left-hand panels), and is also observed in Fig FIGREF42 (the top charts)."}
{"text": "Our Framework ::: Entity-Consistency Augmented Decoder ::: Decoder with Retrieved Entity\tAfter the row selection and column selection, we can define the final retrieved KB entity score as the element-wise dot between the row retriever result and the column selection score, which can be calculated as where the $v^{t}$ indicates the final KB retrieved entity score. Finally, we follow eric:2017:SIGDial to use copy mechanism to incorporate the retrieved entity, which can be defined as where $\\mathbf {o}_t$\u2019s dimensionality is $ |\\mathcal {V}|$ +$|\\mathcal {E}|$. In $\\mathbf {v}^t$ , lower $ |\\mathcal {V}|$ is zero and the rest$|\\mathcal {E}|$ is retrieved entity scores."}
{"text": "12)  Under assumption $\\sigma \\approx 0$ sampling the row of weights from such posterior means putting all the weights from this row either to 0 (drop the corresponding input neuron) or to some learned values. Thus this model is a probabilistic analog of binary dropout with dropout rates $p^x$ and $p^h$ . After unfolding the recurrence in the network, the maximization of the variational lower bound for such model looks as follows:  $$\\sum _{i=1}^\\ell \\int q(\\omega |M) \\log \\Bigl (y^i\\big |f_y\\bigl (f_h(x^i_T, f_h(\\dots f_h (x^i_1, h^i_0))\\bigr )\\Bigr ) d \\omega - \\\\"}
{"text": "This suggests that the methods with better sample efficiency are needed to act in more realistic problem settings with huge action spaces. Experiments also show that a larger memory slot size always helps. Intuitively, with a memory mechanism (either implicit or explicit), an agent could make the environment closer to fully observed by exploring and memorizing observations. Presumably, a larger memory may further improve QA-DQN's performance, but considering the average number of sentences in each iSQuAD game is 5, a memory with more than 5 slots will defeat the purpose of our study of partially observable text environments. Not surprisingly, QA-DQN performs worse in general on iNewsQA, in all experiments. As shown in Table TABREF16, the average number of sentences per document in iNewsQA is about 6 times more than in iSQuAD. This is analogous to games with larger maps in the RL literature, where the environment is partially observable. A better exploration (in our case, jumping) strategy may help QA-DQN to master such harder games."}
{"text": "\\mathbf {o}_t &= \\sigma (\\mathbf {W}_o \\mathbf {x}_t + \\mathbf {U}_o^L \\mathbf {h}_{t-1}^L + \\mathbf {U}_o^R \\mathbf {h}_{t-1}^R),"}
{"text": "Collision Study\tBefore diving into the actual collision studies, it is important to understand what the properties of good projections are. For instance, good projections should be as separate as possible, while still capturing the inherent n-gram features. Words with similar character n-gram feature vectors should be closer to each other i.e. cat and cats, but yet separate from each other so that the network can learn that cat and cats are related, but yet different. Such observations are not evident from the projections. One way to understand them is by looking at the collision rates. For instance, if there are too many projection collisions, this means that the network is fundamentally incapable of learning and it will not be able to generalize. For the purpose, we test how spread out the projections are for word and sentence representations. We take a large corpus enwik9 and analyze the average hamming distance of the words and sentences in the corpus. Intuitively, good projections should have less collisions. Our study shows that there is almost no collision. On an average the Hamming distances between words are 557 bits, which is around 50% of the projection dimension."}
{"text": "ERGM's concept of chance is based on the number of expected pair-wise co-votes between MEPs belonging to a group on a given RCV, knowing the network-related properties of the co-voting network on that particular RCV. The main difference between INLINEFORM2 and ERGM, though, is the treatment of non-voting and abstained MEPs. INLINEFORM3 considers only the yes/no votes, and consequently, agreements by the voting MEPs of the same groups are considerably higher than co-voting by chance. ERGM, on the other hand, always considers all MEPs, and non-voting and abstained MEPs are treated as disconnected nodes. The level of co-voting by chance is therefore considerably lower, since there is often a large fraction of MEPs that do not attand or abstain. As with INLINEFORM0 , Greens-EFA, S&D, and EPP exhibit the highest cohesion, even though their ranking is permuted when compared to the ranking obtained with INLINEFORM1 . At the other end of the scale, we observe the same situation as with INLINEFORM2 . The non-aligned members NI have the lowest cohesion, followed by EFDD and ENL."}
{"text": "Our second contribution is a copy mechanism which allows us to use much smaller decoder vocabulary sizes resulting in much faster decoding. Our copy mechanism also allows us to construct a better representation of out-of-vocabulary words. We demonstrate the effectiveness of our approach in the challenging Gigaword dataset and DUC competition showing state-of-the-art performance."}
{"text": "Experiments ::: Effectiveness of Paired Common Words\tWe further analyse the necessity and effectiveness of modifying sentences with paired common words. We consider another version that replaces one single word independently at each step without using paired common words, namely the unpaired version. Firstly, for matched adversarial examples that can be semantically different from original sentences, the unpaired version is inapplicable, because the matched label can be easily broken if common words from two sentences are changed into other words independently. And for the unmatched case, we show that the unpaired version is much less effective. For a more fair comparison, we double the step number limit for the unpaired version. As shown in Table TABREF41, the performance of target models on unmatched examples generated by the unpaired version, particularly that of BERT, is mostly much higher than those by our full algorithm, except for BiMPM on MRPC but its accuracies have almost reached 0 (0.0% for unpaired and 0.2% for paired). This demonstrates that our algorithm using paired common words are more effective in generating adversarial examples, on which the performance of the target model is generally much lower."}
{"text": "Here we consider five baselines to compare with GraLap: (i) Uniform: assign 3 to all the references assuming equal intensity, (ii) SVR+W: recently proposed Support Vector Regression (SVR) with the feature set mentioned in  , (iii) SVR+O: SVR model with our feature set, (iv) C4.5SSL: C4.5 semi-supervised algorithm with our feature set  , and (v) GLM: the traditional graph-based LP model with our feature set  . Three metrics are used to compare the results of the competing models with the annotated labels: Root Mean Square Error (RMSE), Pearson's correlation coefficient ( INLINEFORM0 ), and coefficient of determination ( INLINEFORM1 ). Table TABREF28 shows the performance of the competing models. We incrementally include each feature set into GraLap greedily on the basis of ranking shown in Figure FIGREF27 (a). We observe that GraLap with only FF outperforms SVR+O with 41% improvement of INLINEFORM0 . As expected, the inclusion of PF into the model improves the model marginally."}
{"text": "In this work we explore the automatic processing of ultrasound tongue images in order to assist SLTs, who currently largely rely on manual processing when using articulatory imaging in speech therapy. One task that could assist SLTs is the automatic classification of tongue shapes from raw ultrasound. This can facilitate the diagnosis and treatment of speech sound disorders, by allowing SLTs to automatically identify incorrect articulations, or by quantifying patient progress in therapy. In addition to being directly useful for speech therapy, the classification of tongue shapes enables further understanding of phonetic variability in ultrasound tongue images. Much of the previous work in this area has focused on speaker-dependent models. In this work we investigate how automatic processing of ultrasound tongue imaging is affected by speaker variation, and how severe degradations in performance can be avoided when applying systems to data from previously unseen speakers through the use of speaker adaptation and speaker normalization approaches. Below, we present the main challenges associated with the automatic processing of ultrasound data, together with a review of speaker-independent models applied to UTI. Following this, we present the experiments that we have performed (Section SECREF2 ), and discuss the results obtained (Section SECREF3 )."}
{"text": "Results\tIn this section, we empirically compare automatic differentiation (AD, our implementation based on Clad) and numerical differentiation (ND, based on finite difference method) in ROOT. We show that AD can drastically improve accuracy and performance of derivative evaluation, compared to ND."}
{"text": "The incidents are wide ranging: from patient accidents to lost forms or referrals; from delays in admission or discharge to serious untoward incidents, such as retained foreign objects after operations. The review and analysis of such data provides critical insight into complex processes in healthcare with a view towards service improvement. Although statistical analyses are routinely performed on the structured data (dates, locations, hand-coded categories, etc), free text is typically read manually and often ignored in practice, unless a detailed review of a case is undertaken because of the severity of harm that resulted. These limitations are due to a lack of methodologies that can provide content-based groupings across the large volume of reports submitted nationally for organisational learning. Automatic categorisation of incidents from free text would sidestep human error and difficulties in assigning incidents to a priori pre-defined lists in the reporting system. Such tools can also offer unbiased insight into the root cause analysis of incidents that could improve the safety and quality of care and efficiency of healthcare services. In this work, we showcase an algorithmic methodology that detects content-based groups of records in an unsupervised manner, based only on the free (unstructured) textual descriptions of the incidents."}
{"text": "Experiments ::: Baselines ::: HSSC @!START@@!END@.\tThis model adopts encoder parameter sharing for jointly sentiment classification and summarization. It predicts the sentiment label using a highway layer, concatenating the hidden state in summary decoder and the original text representation in encoder."}
{"text": "And finally, by comparing $\\text{CMD}^{\\dagger \\dagger }$ with $\\text{CMD}$, or comparing $\\text{DANN}^{\\dagger \\dagger }$ with $\\text{DANN}$, we can know the general effectiveness of our proposed solution."}
{"text": "Precisely, the Softword technique augments the representation of a character with the embedding of its corresponding segmentation label: Here, $seg(c_j) \\in \\mathcal {Y}_{seg}$ denotes the segmentation label of the character $c_j$ predicted by the word segmentor, $\\mathbf {e}^{seg}$ denotes the segmentation label embedding lookup table, and commonly $\\mathcal {Y}_{seg}=\\lbrace \\text{B}, \\text{M}, \\text{E}, \\text{S}\\rbrace $ with B, M, E indicating that the character is the beginning, middle, and end of a word, respectively, and S indicating that the character itself forms a single-character word. The first idea we come out based on the Softword technique is to construct a word segmenter using the lexicon and allow a character to have multiple segmentation labels. Take the sentence $s=\\lbrace c_1, c_2, c_3, c_4, c_5\\rbrace $ as an example."}
{"text": "We also relabeled participants and events from the first annotation phase with ScrEv_other and ScrPart_other, if they did not meet the frequency requirements. The event label air_bathroom (the event of letting fresh air into the room after the bath), for example, was only used once in the stories, so we relabeled that instance to ScrEv_other. Additionally, we looked at the DeScript corpus  , which contains manually clustered event paraphrase sets for the 10 scenarios that are also covered by InScript (see Section \"Comparison to the DeScript Corpus\" ). Every such set contains event descriptions that describe a certain event type. We extended our templates with additional labels for these events, if they were not yet part of the template."}
{"text": "Results and Discussion\tWe present the results in Table TABREF20 . In Zhang et al. zhang2015character it was noted that the performance of ngram features degrades for larger datasets. However, we have seen in our baseline experiments that this effect can be reduced by using suitable frequency cut-offs. We have also seen that in many cases, the ngram features can solely outperform the neural approaches. For the two smaller datasets, linear classifiers tend to perform better, while for the larger datasets performance increases with increasing the non-linear layers even for hand-crafted representations. Combining hand-crafted and learned features is often beneficial, but not always, especially with the linear classifier. What is clear is that different datasets benefit from different representations and model parameters and it is difficult to find a representation that consistently performs well across all datasets. This necessitates repeated experimentation to understand which approaches and parameters would provide more consistent improvements."}
{"text": "Document Visualization\tWord2vec embeddings are known to exhibit linear regularities representing semantic relationships between words  . We explore if these regularities can be transferred to a document representation, when using as a document vector a linear combination of word2vec embeddings. As a weighting scheme we employ LRP or SA scores, with the classifier's predicted class as the target class for the relevance estimation. For comparison we perform uniform weighting, where we simply sum up the word embeddings of the document words (SUM). For SA we use either the $l_2$ -norm or squared $l_2$ -norm for pooling word gradient values along the word2vec dimensions, i.e. in addition to the standard SA word relevance defined in Eq. 15 , we use as an alternative $R_{\\mathrm {SA}(l_2)}(w_t) = \\Vert  \\nabla _{{w_t}} f({d}) \\Vert _2$ and denote this relevance model by SA $(l_2)$ . For both LRP and SA, we employ different variations of the weighting scheme."}
{"text": "Here, $S$ could be used as gold standard for the back-translated paragraph $S_{T_s}$. In this way, we could turn the unsupervised translation into a supervised task by maximizing the similarity between $S$ and $S_{T_S}$. The same also applies to using poem $T$ as gold standard for its corresponding back-translation $T_{S_T}$. We define the following loss: Note that $\\mathcal {L}^{bt}$ does not back propagate through the generation of $T_S$ and $S_T$ as we observe no improvement in doing so. When training the model, we minimize the composite loss: where $\\alpha _1$ and $\\alpha _2$ are scaling factors."}
{"text": "We build an evidence extraction model to predict the most important sub-spans from the passages as evidence, and then develop an answer synthesis model which takes the evidence as additional features along with the question and passage to further elaborate the final answers. Specifically, we develop the answer extraction model with state-of-the-art attention based neural networks which predict the start and end positions of evidence snippets. As multiple passages are provided for each question in the MS-MARCO dataset, we propose incorporating passage ranking as an additional task to improve the results of evidence extraction under a multi-task learning framework. We use the bidirectional recurrent neural networks (RNN) for the word-level representation, and then apply the attention mechanism  to incorporate matching information from question to passage at the word level. Next, we predict start and end positions of the evidence snippet by pointer networks  . Moreover, we aggregate the word-level matching information of each passage using the attention pooling, and use the passage-level representation to rank all candidate passages as an additional task. For the answer synthesis, we apply the sequence-to-sequence model to synthesize the final answer based on the extracted evidence."}
{"text": "Second, we remove from the text of the tweets user names, links, punctuation, tabs, leading and lagging blanks, general spaces and \u201cRT\" - the text that points that a tweet is in fact a retweet. As shown in previous works, emojis are correlated with sentiment . Moreover, as we think that communities will express different sentiment during discussion, it is forseeable that emojis will play an important role as separators of tweets that differentiate between the two sides. Accordingly, we decide to add them to the train-set by translating each emoji into a different word. For example, the emoji :) will be translated into happy and :( into sad. Relations between emojis and words are defined in the R library textclean. Finally, we group tweets by user concatenating them in one string and labeling them with the user's community, namely with tags C1 and C2, corresponding respectively to the biggest and second biggest groups. It is important to note that we take the same number of users of each community to prevent bias in the model. Thus, we use the number of users of the smallest principal community. The train-set built that way is used to feed the model."}
{"text": "QAGS ::: The QAGS Score\tGiven these components, we obtain the QAGS score of a generation by (1) generating $K$ questions conditioned on the summary, (2) answering the questions using both the source article and the summary to get two sets of answers, (3) comparing corresponding answers using the answer similarity metric, and (4) averaging the answer similarity metric over all questions. We depict this process in Figure FIGREF3."}
{"text": "Granularity size is mentioned in terms of durations of event actions  and granularity levels of event participants, time and locations. Granularity is given in terms of partonomic relations or through the part-of relation, between entities and events, using the taxonomy of meronymic relations by  . Granularity levels of the human participant slot are contained within Winston\u00d5s et al. Member-Collection relations. The temporal granularity levels make part of Winston\u00d5s Portion-Mass relationships and locational levels are in line with Place-Area relations in Winston\u00d5s taxonomy. Cybulska and Vossen experimented with a decision-tree supervised pairwise binary classifier to determine coreference of pairs of event mentions. They also ran experiments with a linear SVM and a multinomial Naive Bayes classifier but the decision-tree classifier outperformed both of them. For the experiments, Cybulska and Vossen use the ECB+ dataset  . The ECB+ corpus contains a new corpus component, consisting of 502 texts, describing different instances of event types."}
{"text": "Methodology ::: Machine Learning Models ::: Multi-label Classification\tRAkEL (RAndom k labELsets): RAkEL  is a multi-label classification algorithm that uses labeled powerset (LP) transformation: it basically creates a single binary classifier for every label combination and then uses multiple LP classifiers, each trained on a random subset of the actual labels, for classification."}
{"text": "Method ::: Defining bias in embeddings\tBias can be thought of as \u201cprejudice in favor or against a person, group, or thing that is considered to be unfair\" . Because definitions of fairness have changed over time, algorithms which are trained on \u201creal-world\" data may pick up associations which existed historically (or still exist), but which are considered undesirable. In the word embedding literature, one common idea is to analyse relationships which embeddings encode between professions and gender, race, ethnicity or nationality. We follow this approach in this paper, though note that our method is equally applicable to measuring the encoded relationship between any set of entities in a KG.."}
{"text": "The mechanism merge each spatial location INLINEFORM0 in the annotation sequence INLINEFORM1 with the initial decoder state INLINEFORM2 obtained in equation EQREF7 with non-linearity : DISPLAYFORM0   where INLINEFORM0 is INLINEFORM1 function. The new annotations go through a L2 normalization layer followed by two INLINEFORM2 convolutional layers (of size INLINEFORM3 respectively) to obtain INLINEFORM4 weights, one for each spatial location. We normalize the weights with a softmax to obtain a soft attention map INLINEFORM5 . Each annotation INLINEFORM6 is then weighted according to its corresponding INLINEFORM7 : DISPLAYFORM0   This method can be seen as the removal of unnecessary information in the image annotations according to the source sentence. This attention is used on top of the others - before decoding - and is referred as \"grounded image\" in Table TABREF41 ."}
{"text": "3)  Note that if we want the RDSN to be causal, we can simplify it to  $$P(y_t | x_t, x_{t-1}, ..., x_{t-k}, y_t, y_{t-1}, ..., y_{t-k})$$   (Eq. 4)  where $k$ 's in the above formula represent the number of recurrent frames. Figure 1 shows the framework of RDSN. Adding $y_t, y_{t-1}, ..., y_{t-k}$ as additional inputs, we transferred the pure acoustic model into a hybrid model consisted of an acoustic model and a phoneme-level N-gram model representing the relation between current phone and previous phones. The phoneme-level N-gram (or as in the formula above, k-gram) model provides additional information on phoneme-level, trying to make the output of current frame more accurate and robust with respect to noises and reverberations."}