_id stringlengths 4 10 | text stringlengths 0 18.4k | title stringlengths 0 8.56k |
|---|---|---|
d251765463 | Research on Automatic Story Generation (ASG) relies heavily on human and automatic evaluation. However, there is no consensus on which human evaluation criteria to use, and no analysis of how well automatic criteria correlate with them. In this paper, we propose to re-evaluate ASG evaluation. We introduce a set of 6 orthogonal and comprehensive human criteria, carefully motivated by the social sciences literature. We also present HANNA, an annotated dataset of 1,056 stories produced by 10 different ASG systems. HANNA allows us to quantitatively evaluate the correlations of 72 automatic metrics with human criteria. Our analysis highlights the weaknesses of current metrics for ASG and allows us to formulate practical recommendations for ASG evaluation. | Of Human Criteria and Automatic Metrics: A Benchmark of the Evaluation of Story Generation |
d249017865 | Despite their strong performance on many tasks, pre-trained language models have been shown to struggle on out-of-distribution compositional generalization. Meanwhile, recent work has shown considerable improvements on many NLP tasks from model scaling. Can scaling up model size also improve compositional generalization in semantic parsing? We evaluate encoder-decoder models up to 11B parameters and decoder-only models up to 540B parameters, and compare model scaling curves for three different methods for applying a pre-trained language model to a new task: fine-tuning all parameters, prompt tuning, and in-context learning. We observe that fine-tuning generally has flat or negative scaling curves on out-of-distribution compositional generalization in semantic parsing evaluations. In-context learning has positive scaling curves, but is generally outperformed by much smaller fine-tuned models. Prompt-tuning can outperform fine-tuning, suggesting further potential improvements from scaling as it exhibits a more positive scaling curve. Additionally, we identify several error trends that vary with model scale. For example, larger models are generally better at modeling the syntax of the output space, but are also more prone to certain types of overfitting. Overall, our study highlights limitations of current techniques for effectively leveraging model scale for compositional generalization, while our analysis also suggests promising directions for future work. | Evaluating the Impact of Model Scale for Compositional Generalization in Semantic Parsing |
d259376495 | Reinforcement Learning remains an underutilized method of training and fine-tuning Language Models (LMs) despite recent successes. This paper presents a simple approach of finetuning a language model with Reinforcement Learning to achieve competitive performance on the BEA 2023 Shared Task whose goal is to automatically generate teacher responses in educational dialogues. We utilized the novel NLPO algorithm that masks out tokens during generation to direct the model towards generations that maximize a reward function. We show results for both the t5-base model with 220 million parameters from the HuggingFace repository submitted to the leaderboard that, despite its comparatively small size, has achieved a good performance on both test and dev set, as well as GPT-2 with 124 million parameters. The presented results show that despite maximizing only one of the metrics used in the evaluation as a reward function our model scores highly in the other metrics as well. | Enhancing Educational Dialogues: A Reinforcement Learning Approach for Generating AI Teacher Responses |
d259376594 | This paper describes SinaAI's participation in SemEval-2023 Task 3, which involves detecting propaganda in news articles across multiple languages. The task comprises three subtasks: (i) genre detection, (ii) news framing, and (iii) persuasion technique identification. The employed dataset includes news articles in nine languages and domains, including English, French, Italian, German, Polish, Russian, Georgian, Greek, and Spanish, with labeled instances of news framing, genre, and persuasion techniques. Our approach combines fine-tuning multilingual language models such as XLM, LaBSE, and mBERT with data augmentation techniques. Our experimental results show that XLM outperforms other models in terms of F1-Micro in and F1-Macro, and the ensemble of XLM and LaBSE achieved the best performance. Our study highlights the effectiveness of multilingual sentence embedding models in multilingual propaganda detection. Our models achieved highest score for two languages (Greek and Italian) in sub-task 1 and one language (Russian) for sub-task 2. | SinaAI at SemEval-2023 Task 3: A Multilingual Transformer Language Model-based Approach for the Detection of News Genre, Framing and Persuasion Techniques |
d14347729 | We propose a novel phrase extraction system to generate a phrase dictionary for predictive input methods from a large corpus. This system extracts phrases after counting n-grams so that it can be easily maintained, tuned, and re-executed independently. We developed a rule-based filter based on part-of-speech (POS) patterns to extract Japanese phrases. Our experiment shows usefulness of our system, which achieved a precision of 0.90 and a recall of 0.81, outperforming the N-gram baseline by a large margin. | Phrase Extraction for Japanese Predictive Input Method as Post-Processing |
d7540507 | Previous studies in data-driven dependency parsing have shown that tree transformations can improve parsing accuracy for specific parsers and data sets. We investigate to what extent this can be generalized across languages/treebanks and parsers, focusing on pseudo-projective parsing, as a way of capturing non-projective dependencies, and transformations used to facilitate parsing of coordinate structures and verb groups. The results indicate that the beneficial effect of pseudo-projective parsing is independent of parsing strategy but sensitive to language or treebank specific properties. By contrast, the construction specific transformations appear to be more sensitive to parsing strategy but have a constant positive effect over several languages. | Generalizing Tree Transformations for Inductive Dependency Parsing |
d222291768 | Identifying metaphors in text is very challenging and requires comprehending the underlying comparison. The automation of this cognitive process has gained wide attention lately. However, the majority of existing approaches concentrate on word-level identification by treating the task as either single-word classification or sequential labelling without explicitly modelling the interaction between the metaphor components. On the other hand, while existing relation-level approaches implicitly model this interaction, they ignore the context where the metaphor occurs. In this work, we address these limitations by introducing a novel architecture for identifying relation-level metaphoric expressions of certain grammatical relations based on contextual modulation. In a methodology inspired by works in visual reasoning, our approach is based on conditioning the neural network computation on the deep contextualised features of the candidate expressions using feature-wise linear modulation. We demonstrate that the proposed architecture achieves state-of-the-art results on benchmark datasets. The proposed methodology is generic and could be applied to other textual classification problems that benefit from contextual interaction. | Contextual Modulation for Relation-Level Metaphor Identification |
d259376755 | This paper describes the speech translation system submitted as part of the IWSLT 2023 shared task on low resource speech translation. The low resource task aids in building models for language pairs where the training corpus is limited. In this paper, we focus on two language pairs, namely, Tamasheq-French (Tmh→Fra) and Marathi-Hindi (Mr→Hi) and implement a speech translation system that is unconstrained. We evaluate three strategies in our system: (a) Data augmentation where we perform different operations on audio as well as text samples, (b) an ensemble model that integrates a set of models trained using a combination of augmentation strategies, and (c) post-processing techniques where we explore the use of large language models (LLMs) to improve the quality of sentences that are generated. Experiments show how data augmentation can relatively improve the BLEU score by 5.2% over the baseline system for Tmh→Fra while an ensemble model further improves performance by 17% for Tmh→Fra and 23% for Mr→Hi task. . 2020. wav2vec 2.0: A framework for self-supervised learning of speech representations. | Improving Low Resource Speech Translation with Data Augmentation and Ensemble Strategies |
d258378286 | This paper presents the development of the AIbased language-learning platform, Revita. It is an intelligent online tutor, developed to support learners of multiple languages, from lowerintermediate toward advanced levels. It has been in pilot use with hundreds of students at several universities, whose feedback and needs shape the development. One of the main emerging features of Revita is the system of linguistic constructs to represent the domain knowledge. The system of constructs is developed in collaboration with experts in language pedagogy. Constructs define the types of exercises, the content of the feedback, and enable detailed modeling and evaluation of learner progress. | Linguistic Constructs Represent the Domain Model in Intelligent Language Tutoring |
d252595790 | For most natural language processing tasks, the dominant practice is to finetune large pretrained transformer models (e.g., BERT) using smaller downstream datasets. Despite the success of this approach, it remains unclear to what extent these gains are attributable to the massive background corpora employed for pretraining versus to the pretraining objectives themselves. This paper introduces a large-scale study of self-pretraining, where the same (downstream) training data is used for both pretraining and finetuning. In experiments addressing both ELECTRA and RoBERTa models and 10 distinct downstream classification datasets, we observe that self-pretraining rivals standard pretraining on the BookWiki corpus (despite using around 10×-500× less data), outperforming the latter on 7 and 5 datasets, respectively. Surprisingly, these task-specific pretrained models often perform well on other tasks, including the GLUE benchmark. Self-pretraining also provides benefits on structured output prediction tasks such as question answering and commonsense inference, often providing more than 50% improvements compared to standard pretraining. Our results hint that often performance gains attributable to pretraining are driven primarily by the pretraining objective itself and are not always attributable to the use of external pretraining data in massive amounts. These findings are especially relevant in light of concerns about intellectual property and offensive content in web-scale pretraining data. 1 | Downstream Datasets Make Surprisingly Good Pretraining Corpora |
d21686919 | What makes some types of languages more probable than others? For instance, we know that almost all spoken languages contain the vowel phoneme /i/; why should that be? The field of linguistic typology seeks to answer these questions and, thereby, divine the mechanisms that underlie human language. In our work, we tackle the problem of vowel system typology, i.e., we propose a generative probability model of which vowels a language contains. In contrast to previous work, we work directly with the acoustic information-the first two formant values-rather than modeling discrete sets of phonemic symbols (IPA). We develop a novel generative probability model and report results based on a corpus of 233 languages. | A Deep Generative Model of Vowel Formant Typology |
d218595134 | This paper studies strategies to model word formation in NMT using rich linguistic information, namely a word segmentation approach that goes beyond splitting into substrings by considering fusional morphology. Our linguistically sound segmentation is combined with a method for target-side inflection to accommodate modeling word formation. The best system variants employ source-side morphological analysis and model complex target-side words, improving over a standard system. | |
d259376756 | We describe our submission to SemEval 2023 Task 3, specifically the subtask on persuasion technique detection. In this work, our team tackled a novel task of classifying persuasion techniques in online news articles at a paragraph level. The low-resource multilingual datasets, along with the imbalanced label distribution, make this task challenging. Our team presented a cross-lingual data augmentation approach and leveraged a recently proposed multilingual natural language inference model to address these challenges. Our solution achieves the highest macro-F1 score for the English task, and top 5 micro-F1 scores on both the English and Russian leaderboards. We have made the source code of our models and experiments publically available at 1 . | NLUBot101 at SemEval-2023 Task 3: An Augmented Multilingual NLI Approach Towards Online News Persuasion Techniques Detection |
d218684889 | Exploiting natural language processing in the clinical domain requires de-identification, i.e., anonymization of personal information in texts. However, current research considers de-identification and downstream tasks, such as concept extraction, only in isolation and does not study the effects of de-identification on other tasks. In this paper, we close this gap by reporting concept extraction performance on automatically anonymized data and investigating joint models for de-identification and concept extraction. In particular, we propose a stacked model with restricted access to privacy-sensitive information and a multitask model. We set the new state of the art on benchmark datasets in English (96.1% F1 for de-identification and 88.9% F1 for concept extraction) and Spanish (91.4% F1 for concept extraction).PATIENT AGE PROBLEMPatrick died of myocardial infarction (MI) at 76. | Closing the Gap: Joint De-Identification and Concept Extraction in the Clinical Domain |
d259370766 | In this paper, we investigate the potential of using large pre-trained language models to generate non-playable character (NPC) scripts in video games. We introduce a novel pipeline that automatically constructs believable NPC scripts for various game genres and specifications using Transformer-based models. Moreover, we develop a self-diagnosis method, inspired by prior research, that is tailored to essential NPC characteristics such as coherence, believability, and variety in dialogue. To evaluate our approach, we propose a new benchmark, The Turing Quest, which demonstrates that our pipeline, when applied to GPT-3, generates NPC scripts across diverse game genres and contexts that can successfully deceive judges into believing they were written by humans. Our findings hold significant implications for the gaming industry and its global community, as the current reliance on manually-curated scripts is resource-intensive and can limit the immersiveness and enjoyment of players. | The Turing Quest: Can Transformers Make Good NPCs? |
d259370803 | Asking and answering questions are inseparable parts of human social life. The primary purposes of asking questions are to gain knowledge or request help which has been the subject of question-answering studies. However, questions can also reflect negative intentions and include implicit offenses, such as highlighting one's lack of knowledge or bolstering an alleged superior knowledge, which can lead to conflict in conversations; yet has been scarcely researched. This paper is the first study to introduce a dataset (Question Intention Dataset) that includes questions with positive/neutral and negative intentions and the underlying intention categories within each group. We further conduct a meta-analysis to highlight tacit and apparent intents. We also propose a classification method using Transformers augmented by TF-IDF-based features and report the results of several models for classifying the main intention categories. We aim to highlight the importance of taking intentions into account, especially implicit and negative ones, to gain insight into conflict-evoking questions and better understand human-human communication on the web for NLP applications. | What is the Real Intention behind this Question? Dataset Collection and Intention Classification |
d253628241 | With Web API technology becoming increasingly mature, how to integrate Web API and Chatbot technology has become an issue of great interest. This study plans to build a semi-automatic method and tool, BOTEN. This method allows application developers to build Chatbot interfaces with specified Web APIs quickly. To ensure that the Chatbot has sufficient natural language understanding (NLU) capability, this research evaluates the training sentences written by the developer through TF-IDF, WordNet, and SpaCy techniques, and suggests the developer modify the training sentences with poor quality. This technique can also be used to automatically increase the number of training sentences to improve the capability of Intent recognition.關鍵字：對 話 機 器 人 、Web API、Rasa、 WordNet、SpaCy | |
d250179891 | Nous présentons ici FENEC (FrEnch Named-entity Evaluation Corpus), un corpus à échantillons équilibrés contenant six genres, annoté en entités nommées selon le schéma fin Quaero. Les caractéristiques de ce corpus nous permettent d'évaluer et de comparer trois outils d'annotation automatique -un à base de règles et deux à base de réseaux de neurones -en jouant sur trois dimensions : la finesse du jeu d'étiquettes, le genre des corpus, et les métriques d'évaluation.ABSTRACTFENEC : a balanced sample corpus for French named entity recognitionWe present FENEC (FrEnch Named-entity Evaluation Corpus), a balanced sample corpus containing six genres and annotated with named entities according to Quaero, a rich annotation scheme. The characteristics of this corpus allow us to evaluate and compare three automatic annotation tools-one rule-based and two neural network-based-by playing on three dimensions of the evaluation: the precision of the label set, the genre of the corpora, and the evaluation metrics. | |
d2279432 | We address the problem of analyzing multiple related opinions in a text. For instance, in a restaurant review such opinions may include food, ambience and service. We formulate this task as a multiple aspect ranking problem, where the goal is to produce a set of numerical scores, one for each aspect. We present an algorithm that jointly learns ranking models for individual aspects by modeling the dependencies between assigned ranks. This algorithm guides the prediction of individual rankers by analyzing meta-relations between opinions, such as agreement and contrast. We prove that our agreementbased joint model is more expressive than individual ranking models. Our empirical results further confirm the strength of the model: the algorithm provides significant improvement over both individual rankers and a state-of-the-art joint ranking model. | Multiple Aspect Ranking using the Good Grief Algorithm |
d2340513 | The availability of large scale data sets of manually annotated predicate-argument structures has recently favored the use of machine learning approaches to the design of automated semantic role labeling (SRL) systems. The main research in this area relates to the design choices for feature representation and for effective decompositions of the task in different learning models. Regarding the former choice, structural properties of full syntactic parses are largely employed as they represent ways to encode different principles suggested by the linking theory between syntax and semantics. The latter choice relates to several learning schemes over global views of the parses. For example, re-ranking stages operating over alternative predicate-argument sequences of the same sentence have shown to be very effective.In this article, we propose several kernel functions to model parse tree properties in kernelbased machines, for example, perceptrons or support vector machines. In particular, we define different kinds of tree kernels as general approaches to feature engineering in SRL. Moreover, we extensively experiment with such kernels to investigate their contribution to individual stages of an SRL architecture both in isolation and in combination with other traditional manually coded features. The results for boundary recognition, classification, and re-ranking stages provide systematic evidence about the significant impact of tree kernels on the overall accuracy, especially when the amount of training data is small. As a conclusive result, tree kernels allow for a general and easily portable feature engineering method which is applicable to a large family of natural language processing tasks. | Tree Kernels for Semantic Role Labeling |
d16708497 | Decoding of phrase-based translation models in the general case is known to be NPcomplete, by a reduction from the traveling salesman problem(Knight, 1999). In practice, phrase-based systems often impose a hard distortion limit that limits the movement of phrases during translation. However, the impact on complexity after imposing such a constraint is not well studied. In this paper, we describe a dynamic programming algorithm for phrase-based decoding with a fixed distortion limit. The runtime of the algorithm is O(nd!lh d+1 ) where n is the sentence length, d is the distortion limit, l is a bound on the number of phrases starting at any position in the sentence, and h is related to the maximum number of target language translations for any source word. The algorithm makes use of a novel representation that gives a new perspective on decoding of phrase-based models.59 | A Polynomial-Time Dynamic Programming Algorithm for Phrase-Based Decoding with a Fixed Distortion Limit |
d2633377 | Descriptive document clustering aims to automatically discover groups of semantically related documents and to assign a meaningful label to characterise the content of each cluster. In this paper, we present a descriptive clustering approach that employs a distributed representation model, namely the paragraph vector model, to capture semantic similarities between documents and phrases. The proposed method uses a joint representation of phrases and documents (i.e., a coembedding) to automatically select a descriptive phrase that best represents each document cluster. We evaluate our method by comparing its performance to an existing state-of-the-art descriptive clustering method that also uses co-embedding but relies on a bag-of-words representation. Results obtained on benchmark datasets demonstrate that the paragraph vector-based method obtains superior performance over the existing approach in both identifying clusters and assigning appropriate descriptive labels to them. | Distributed Document and Phrase Co-embeddings for Descriptive Clustering |
d21709048 | We present the IIT Bombay English-Hindi Parallel Corpus. The corpus is a compilation of parallel corpora previously available in the public domain as well as new parallel corpora we collected. The corpus contains 1.49 million parallel segments, of which 694k segments were not previously available in the public domain. The corpus has been pre-processed for machine translation, and we report baseline phrase-based SMT and NMT translation results on this corpus. This corpus has been used in two editions of shared tasks at the Workshop on Asian Language Translation (2016 and 2017). The corpus is freely available for non-commercial research. To the best of our knowledge, this is the largest publicly available English-Hindi parallel corpus. | |
d252847494 | Dialogue discourse parsing aims to uncover the internal structure of a multi-participant conversation by finding all the discourse links and corresponding relations. Previous work either treats this task as a series of independent multiple-choice problems, in which the link existence and relations are decoded separately, or the encoding is restricted to only local interaction, ignoring the holistic structural information. In contrast, we propose a principled method that improves upon previous work from two perspectives: encoding and decoding. From the encoding side, we perform structured encoding on the adjacency matrix followed by the matrix-tree learning algorithm, where all discourse links and relations in the dialogue are jointly optimized based on latent tree-level distribution. From the decoding side, we perform structured inference using the modified Chiu-Liu-Edmonds algorithm, which explicitly generates the labeled multi-root non-projective spanning tree that best captures the discourse structure. In addition, unlike in previous work, we do not rely on hand-crafted features; this improves the model's robustness. Experiments show that our method achieves new state-of-the-art, surpassing the previous model by 2.3 on STAC and 1.5 on Molweni (F1 scores). 1 | Structured Dialogue Discourse Parsing |
d7042984 | Punctuations are not available in automatic speech recognition outputs, which could create barriers to many subsequent text processing tasks. This paper proposes a novel method to predict punctuation symbols for the stream of words in transcribed speech texts. Our method jointly performs parsing and punctuation prediction by integrating a rich set of syntactic features when processing words from left to right. It can exploit a global view to capture long-range dependencies for punctuation prediction with linear complexity. The experimental results on the test data sets of IWSLT and TDT4 show that our method can achieve high-level performance in punctuation prediction over the stream of words in transcribed speech text. | Punctuation Prediction with Transition-based Parsing |
d230523780 | We present the first multi-task learning model-named PhoNLP-for joint Vietnamese part-of-speech (POS) tagging, named entity recognition (NER) and dependency parsing. Experiments on Vietnamese benchmark datasets show that PhoNLP produces state-of-the-art results, outperforming a single-task learning approach that fine-tunes the pre-trained Vietnamese language model PhoBERT (Nguyen and Nguyen, 2020) for each task independently. We publicly release PhoNLP as an open-source toolkit under the Apache License 2.0. Although we specify PhoNLP for Vietnamese, our PhoNLP training and evaluation command scripts in fact can directly work for other languages that have a pre-trained BERT-based language model and gold annotated corpora available for the three tasks of POS tagging, NER and dependency parsing. We hope that PhoNLP can serve as a strong baseline and useful toolkit for future NLP research and applications to not only Vietnamese but also the other languages. Our PhoNLP is available at https://github. com/VinAIResearch/PhoNLP. | PhoNLP: A joint multi-task learning model for Vietnamese part-of-speech tagging, named entity recognition and dependency parsing |
d16083742 | We propose a model for knowledge attribution and epistemic evaluation in scientific discourse, consisting of three dimensions with different values: source (author, other, unknown); value (unknown, possible, probable, presumed true) and basis (reasoning, data, other). Based on a literature review, we investigate four linguistic features that mark different types epistemic evaluation (modal auxiliary verbs, adverbs/adjectives, reporting verbs and references). A corpus study on two biology papers indicates the usefulness of this model, and suggest some typical trends. In particular, we find that matrix clauses with a reporting verb of the form 'These results suggest', are the predominant feature indicating knowledge attribution in scientific text. | Epistemic Modality and Knowledge Attribution in Scientific Discourse: A Taxonomy of Types and Overview of Features |
d201058596 | Machine comprehension of texts longer than a single sentence often requires coreference resolution. However, most current reading comprehension benchmarks do not contain complex coreferential phenomena and hence fail to evaluate the ability of models to resolve coreference. We present a new crowdsourced dataset containing more than 24K span-selection questions that require resolving coreference among entities in over 4.7K English paragraphs from Wikipedia. Obtaining questions focused on such phenomena is challenging, because it is hard to avoid lexical cues that shortcut complex reasoning. We deal with this issue by using a strong baseline model as an adversary in the crowdsourcing loop, which helps crowdworkers avoid writing questions with exploitable surface cues. We show that state-of-the-art reading comprehension models perform significantly worse than humans on this benchmark-the best model performance is 70.5 F 1 , while the estimated human performance is 93.4 F 1 . | QUOREF: A Reading Comprehension Dataset with Questions Requiring Coreferential Reasoning |
d222141733 | Multimodal automatic speech recognition systems integrate information from images to improve speech recognition quality, by grounding the speech in the visual context. While visual signals have been shown to be useful for recovering entities that have been masked in the audio, these models should be capable of recovering a broader range of word types. Existing systems rely on global visual features that represent the entire image, but localizing the relevant regions of the image will make it possible to recover a larger set of words, such as adjectives and verbs. In this paper, we propose a model that uses finer-grained visual information from different parts of the image, using automatic object proposals. In experiments on the Flickr8K Audio Captions Corpus, we find that our model improves over approaches that use global visual features, that the proposals enable the model to recover entities and other related words, such as adjectives, and that improvements are due to the model's ability to localize the correct proposals. 1 | Fine-Grained Grounding for Multimodal Speech Recognition |
d252668651 | Being able to train Named Entity Recognition (NER) models for emerging topics is crucial for many real-world applications especially in the medical domain where new topics are continuously evolving out of the scope of existing models and datasets. For a realistic evaluation setup, we introduce a novel COVID-19 news NER dataset (COVIDNEWS-NER) and release 3000 entries of hand annotated strongly labelled sentences and 13000 auto-generated weakly labelled sentences. Besides the dataset, we propose CONTROSTER, a recipe to strategically combine weak and strong labels in improving NER in an emerging topic through transfer learning. We show the effectiveness of CONTROSTER on COVIDNEWS-NER while providing analysis on combining weak and strong labels for training. Our key findings are: (1) Using weak data to formulate an initial backbone before tuning on strong data outperforms methods trained on only strong or weak data. (2) A combination of out-of-domain and in-domain weak label training is crucial and can overcome saturation when being training on weak labels from a single source. 1 | How to tackle an emerging topic? Combining strong and weak labels for Covid news NER |
d2082658 | From the perspective of structural linguistics, we explore paradigmatic and syntagmatic lexical relations for Chinese POS tagging, an important and challenging task for Chinese language processing. Paradigmatic lexical relations are explicitly captured by word clustering on large-scale unlabeled data and are used to design new features to enhance a discriminative tagger. Syntagmatic lexical relations are implicitly captured by constituent parsing and are utilized via system combination. Experiments on the Penn Chinese Treebank demonstrate the importance of both paradigmatic and syntagmatic relations. Our linguistically motivated approaches yield a relative error reduction of 18% in total over a stateof-the-art baseline. * This work is mainly finished when this author (corresponding author) was in Saarland University and DFKI. | Capturing Paradigmatic and Syntagmatic Lexical Relations: Towards Accurate Chinese Part-of-Speech Tagging |
d256460914 | Patronizing and condescending language is characterized, among others, by its subtle nature. It thus seems reasonable to assume that detecting condescending language in text would be harder than detecting more explicitly harmful language, such as hate speech. However, the results of a SemEval-2022 Task devoted to this topic paint a different picture, with the top-performing systems achieving remarkably strong results. In this paper, we analyse the surprising effectiveness of standard text classification methods in more detail. In particular, we highlight the presence of two rather different types of condescending language in the dataset from the SemEval task. Some inputs are condescending because of the way they talk about a particular subject, i.e. condescending language in this case is a linguistic phenomenon, which can, in principle, be learned from training examples. However, other inputs are condescending because of the nature of what is said, rather than the way in which it is expressed, e.g. by emphasizing stereotypes about a given community. In such cases, our ability to detect condescending language, with current methods, largely depends on the presence of similar examples in the training data. | Identifying Condescending Language: A Tale of Two Distinct Phenomena? |
d256461124 | In this paper, we describe our system, Pre-siUniv, to generate and rank candidate simplifications using publicly available pre-trained language models (BERT, BETO, and BERTimbeau), word embeddings (Eg. FastText, NILC), and part-of-speech taggers (NLTK PoS Tagger, Stanford PoS Tagger and Mac-Morpho), to generate and rank candidate contextual simplifications for a given complex word. In this shared task, our system was placed first in the Spanish track, 5th in the Brazilian-Portuguese track, and 10th in the English track. We upload our codes and data for this project to aid in replication of our results. We also analyze some of the errors and describe design decisions which we took while writing the paper. | PresiUniv at TSAR-2022 Shared Task: Generation and Ranking of Simplification Substitutes of Complex Words in Multiple Languages |
d211133237 | For over a decade, TV series have been drawing increasing interest, both from the audience and from various academic fields. But while most viewers are hooked on the continuous plots of TV serials, the few annotated datasets available to researchers focus on standalone episodes of classical TV series. We aim at filling this gap by providing the multimedia/speech processing communities with Serial Speakers, an annotated dataset of 155 episodes from three popular American TV serials: Breaking Bad, Game of Thrones and House of Cards. Serial Speakers is suitable both for investigating multimedia retrieval in realistic use case scenarios, and for addressing lower level speech related tasks in especially challenging conditions. We publicly release annotations for every speech turn (boundaries, speaker) and scene boundary, along with annotations for shot boundaries, recurring shots, and interacting speakers in a subset of episodes. Because of copyright restrictions, the textual content of the speech turns is encrypted in the public version of the dataset, but we provide the users with a simple online tool to recover the plain text from their own subtitle files. | c European Language Resources Association (ELRA), licensed under |
d8443958 | The ability to compute an accurate reward function is essential for optimising a dialogue policy via reinforcement learning. In real-world applications, using explicit user feedback as the reward signal is often unreliable and costly to collect. This problem can be mitigated if the user's intent is known in advance or data is available to pre-train a task success predictor off-line. In practice neither of these apply for most real world applications. Here we propose an on-line learning framework whereby the dialogue policy is jointly trained alongside the reward model via active learning with a Gaussian process model. This Gaussian process operates on a continuous space dialogue representation generated in an unsupervised fashion using a recurrent neural network encoder-decoder. The experimental results demonstrate that the proposed framework is able to significantly reduce data annotation costs and mitigate noisy user feedback in dialogue policy learning. | On-line Active Reward Learning for Policy Optimisation in Spoken Dialogue Systems |
d245144297 | AMR parsing has experienced an unprecendented increase in performance in the last three years, due to a mixture of effects including architecture improvements and transfer learning. Self-learning techniques have also played a role in pushing performance forward. However, for most recent high performant parsers, the effect of self-learning and silver data augmentation seems to be fading. In this paper we propose to overcome this diminishing returns of silver data by combining Smatch-based ensembling techniques with ensemble distillation. In an extensive experimental setup, we push single model English parser performance to a new state-of-the-art, 85.9 (AMR2.0) and 84.3 (AMR3.0), and return to substantial gains from silver data augmentation. We also attain a new state-of-the-art for cross-lingual AMR parsing for Chinese, German, Italian and Spanish. Finally we explore the impact of the proposed technique on domain adaptation, and show that it can produce gains rivaling those of human annotated data for QALD-9 and achieve a new state-of-the-art for BioAMR. | Maximum Bayes Smatch Ensemble Distillation for AMR Parsing |
d226262365 | Policy gradients-based reinforcement learning has proven to be a promising approach for directly optimizing non-differentiable evaluation metrics for language generation tasks. However, optimizing for a specific metric reward leads to improvements in mostly that metric only, suggesting that the model is gaming the formulation of that metric in a particular way without often achieving real qualitative improvements. Hence, it is more beneficial to make the model optimize multiple diverse metric rewards jointly. While appealing, this is challenging because one needs to manually decide the importance and scaling weights of these metric rewards. Further, it is important to consider using a dynamic combination and curriculum of metric rewards that flexibly changes over time. Considering the above aspects, in our work, we automate the optimization of multiple metric rewards simultaneously via a multi-armed bandit approach (DORB), where at each round, the bandit chooses which metric reward to optimize next, based on expected arm gains. We use the Exp3 algorithm for bandits and formulate two approaches for bandit rewards: (1) Single Multi-reward Bandit (SM-Bandit); (2) Hierarchical Multi-reward Bandit (HM-Bandit). We empirically show the effectiveness of our approaches via various automatic metrics and human evaluation on two important NLG tasks: question generation and data-to-text generation, including on an unseen-test transfer setup. Finally, we present interpretable analyses of the learned bandit curriculum over the optimized rewards. | DORB: Dynamically Optimizing Multiple Rewards with Bandits |
d252992654 | Prompt tuning is a new few-shot transfer learning technique that only tunes the learnable prompt for pre-trained vision and language models such as CLIP. However, existing prompt tuning methods tend to learn spurious or entangled representations, which leads to poor generalization to unseen concepts. Towards non-spurious and efficient prompt learning from limited examples, this paper presents a novel Counterfactual Prompt Learning (CPL) method for vision and language models, which simultaneously employs counterfactual generation and contrastive learning in a joint optimization framework. Particularly, CPL constructs counterfactual by identifying minimal non-spurious feature change between semantically-similar positive and negative samples that causes concept change and learns more generalizable prompt representation from both factual and counterfactual examples via contrastive learning. Extensive experiments demonstrate that CPL can obtain superior few-shot performance on different vision and language tasks than previous prompt tuning methods on CLIP. On image classification, we achieve a 3.55% average relative improvement on unseen classes across seven datasets; on image-text retrieval and visual question answering, we gain up to 4.09% and 25.08% relative improvements across three few-shot scenarios on unseen test sets respectively. 1 | CPL: Counterfactual Prompt Learning for Vision and Language Models |
d2423360 | We introduce a neural semantic parser which converts natural language utterances to intermediate representations in the form of predicate-argument structures, which are induced with a transition system and subsequently mapped to target domains. The semantic parser is trained end-to-end using annotated logical forms or their denotations. We achieve the state of the art on SPADES and GRAPHQUESTIONS and obtain competitive results on GEOQUERY and WEB-QUESTIONS.The induced predicateargument structures shed light on the types of representations useful for semantic parsing and how these are different from linguistically motivated ones. 1 | Learning Structured Natural Language Representations for Semantic Parsing |
d219260475 | Hateful rhetoric is plaguing online discourse, fostering extreme societal movements and possibly giving rise to real-world violence. A potential solution to this growing global problem is citizen-generated counter speech where citizens actively engage with hate speech to restore civil non-polarized discourse. However, its actual effectiveness in curbing the spread of hatred is unknown and hard to quantify. One major obstacle to researching this question is a lack of large labeled data sets for training automated classifiers to identify counter speech. Here we use a unique situation in Germany where self-labeling groups engaged in organized online hate and counter speech. We use an ensemble learning algorithm which pairs a variety of paragraph embeddings with regularized logistic regression functions to classify both hate and counter speech in a corpus of millions of relevant tweets from these two groups. Our pipeline achieves macro F1 scores on out of sample balanced test sets ranging from 0.76 to 0.97-accuracy in line and even exceeding the state of the art. We then use the classifier to discover hate and counter speech in more than 135,000 fully-resolved Twitter conversations occurring from 2013 to 2018 and study their frequency and interaction. Altogether, our results highlight the potential of automated methods to evaluate the impact of coordinated counter speech in stabilizing conversations on social media. ⇤ Denotes equal contribution. | Countering hate on social media: Large-scale classification of hate and counter speech |
d6225155 | Language documentation begins by gathering speech. Manual or automatic transcription at the word level is typically not possible because of the absence of an orthography or prior lexicon, and though manual phonemic transcription is possible, it is prohibitively slow. On the other hand, translations of the minority language into a major language are more easily acquired. We propose a method to harness such translations to improve automatic phoneme recognition. The method assumes no prior lexicon or translation model, instead learning them from phoneme lattices and translations of the speech being transcribed. Experiments demonstrate phoneme error rate improvements against two baselines and the model's ability to learn useful bilingual lexical entries. | Learning a Lexicon and Translation Model from Phoneme Lattices |
d248863142 | Recent research on style transfer takes inspiration from unsupervised neural machine translation (UNMT), learning from large amounts of non-parallel data by exploiting cycle consistency loss, back-translation, and denoising autoencoders. By contrast, the use of selfsupervised NMT (SSNMT), which leverages (near) parallel instances hidden in non-parallel data more efficiently than UNMT, has not yet been explored for style transfer. In this paper we present a novel Self-Supervised Style Transfer (3ST) model, which augments SS-NMT with UNMT methods in order to identify and efficiently exploit supervisory signals in non-parallel social media posts. We compare 3ST with state-of-the-art (SOTA) style transfer models across civil rephrasing, formality and polarity tasks. We show that 3ST is able to balance the three major objectives (fluency, content preservation, attribute transfer accuracy) the best, outperforming SOTA models on averaged performance across their tested tasks in automatic and human evaluation. | Exploiting Social Media Content for Self-Supervised Style Transfer |
d52247458 | The problem of entity-typing has been studied predominantly in supervised learning fashion, mostly with task-specific annotations (for coarse types) and sometimes with distant supervision (for fine types). While such approaches have strong performance within datasets, they often lack the flexibility to transfer across text genres and to generalize to new type taxonomies. In this work we propose a zero-shot entity typing approach that requires no annotated data and can flexibly identify newly defined types.Given a type taxonomy defined as Boolean functions of FREEBASE "types", we ground a given mention to a set of type-compatible Wikipedia entries and then infer the target mention's types using an inference algorithm that makes use of the types of these entries. We evaluate our system on a broad range of datasets, including standard fine-grained and coarse-grained entity typing datasets, and also a dataset in the biological domain. Our system is shown to be competitive with state-of-theart supervised NER systems and outperforms them on out-of-domain datasets. We also show that our system significantly outperforms other zero-shot fine typing systems. | Zero-Shot Open Entity Typing as Type-Compatible Grounding |
d3006364 | We present WiBi, an approach to the automatic creation of a bitaxonomy for Wikipedia, that is, an integrated taxonomy of Wikipage pages and categories. We leverage the information available in either one of the taxonomies to reinforce the creation of the other taxonomy. Our experiments show higher quality and coverage than state-of-the-art resources like DBpedia, YAGO, MENTA, WikiNet and WikiTaxonomy. WiBi is available at | Two Is Bigger (and Better) Than One: the Wikipedia Bitaxonomy Project |
d3051526 | Despite being closely related languages, German and English are characterized by important word order differences. Longrange reordering of verbs, in particular, represents a real challenge for state-of-theart SMT systems and is one of the main reasons why translation quality is often so poor in this language pair. In this work, we review several solutions to improve the accuracy of German-English word reordering while preserving the efficiency of phrase-based decoding. Among these, we consider a novel technique to dynamically shape the reordering search space and effectively capture long-range reordering phenomena. Through an extensive evaluation including diverse translation quality metrics, we show that these solutions can significantly narrow the gap between phrase-based and hierarchical SMT. | Efficient solutions for word reordering in German-English phrase-based statistical machine translation |
d258378261 | Writing is an important element of language learning, and an increasing amount of learner writing is taking place in online environments. Teachers can provide valuable feedback by commenting on learner text. However, providing relevant feedback for every issue for every student can be time-consuming. To address this, we turn to the NLP subfield of feedback comment generation, the task of automatically generating explanatory notes for learner text with the goal of enhancing learning outcomes. However, freely-generated comments may mix multiple topics seen in the training data or even give misleading advice. In this thesis proposal, we seek to address these issues by categorizing comments and constraining the outputs of noisy classes. We describe an annotation scheme for feedback comment corpora using comment topics with a broader scope than existing typologies focused on error correction. We outline plans for experiments in grouping and clustering, replacing particularly diverse categories with modular templates, and comparing the generation results of using different linguistic features and model architectures with the original dataset versus the newly annotated one. This paper presents the first two years (the master's component) of a research project for a five-year combined master's and Ph.D program. | Template-guided Grammatical Error Feedback Comment Generation |
d11438927 | Many argumentative texts, and news editorials in particular, follow a specific strategy to persuade their readers of some opinion or attitude. This includes decisions such as when to tell an anecdote or where to support an assumption with statistics, which is reflected by the composition of different types of argumentative discourse units in a text. While several argument mining corpora have recently been published, they do not allow the study of argumentation strategies due to incomplete or coarse-grained unit annotations. This paper presents a novel corpus with 300 editorials from three diverse news portals that provides the basis for mining argumentation strategies. Each unit in all editorials has been assigned one of six types by three annotators with a high Fleiss' κ agreement of 0.56. We investigate various challenges of the annotation process and we conduct a first corpus analysis. Our results reveal different strategies across the news portals, exemplifying the benefit of studying editorials-a so far underresourced text genre in argument mining. | A News Editorial Corpus for Mining Argumentation Strategies |
d258557820 | Jointly extracting entity pairs and their relations is challenging when working on distantlysupervised data with ambiguous or noisy labels. To mitigate such impact, we propose uncertainty-aware bootstrap learning, which is motivated by the intuition that the higher uncertainty of an instance, the more likely the model confidence is inconsistent with the ground truths. Specifically, we first explore instance-level data uncertainty to create an initial high-confident examples. Such subset serves as filtering noisy instances and facilitating the model to converge fast at the early stage. During bootstrap learning, we propose self-ensembling as a regularizer to alleviate inter-model uncertainty produced by noisy labels. We further define probability variance of joint tagging probabilities to estimate innermodel parametric uncertainty, which is used to select and build up new reliable training instances for the next iteration. Experimental results on two large datasets reveal that our approach outperforms existing strong baselines and related methods. | Uncertainty-Aware Bootstrap Learning for Joint Extraction on Distantly-Supervised Data |
d221377163 | Research on definition extraction has been conducted for well over a decade, largely with significant constraints on the type of definitions considered. In this work, we present DeftEval, a SemEval shared task in which participants must extract definitions from free text using a termdefinition pair corpus that reflects the complex reality of definitions in natural language. Definitions and glosses in free text often appear without explicit indicators, across sentences boundaries, or in an otherwise complex linguistic manner. DeftEval involved 3 distinct subtasks: 1) Sentence classification, 2) sequence labeling, and 3) relation extraction.This work is licensed under a Creative Commons Attribution 4.0 International License.License details: | SemEval-2020 Task 6: Definition extraction from free text with the DEFT corpus |
d57189197 | Aspect level sentiment classification is a finegrained sentiment analysis task. To detect the sentiment towards a particular aspect in a sentence, previous studies have developed various attention-based methods for generating aspect-specific sentence representations. However, the attention may inherently introduce noise and downgrade the performance. In this paper, we propose constrained attention networks (CAN), a simple yet effective solution, to regularize the attention for multiaspect sentiment analysis, which alleviates the drawback of the attention mechanism. Specifically, we introduce orthogonal regularization on multiple aspects and sparse regularization on each single aspect. Experimental results on two public datasets demonstrate the effectiveness of our approach. We further extend our approach to multi-task settings and outperform the state-of-the-art methods. | CAN: Constrained Attention Networks for Multi-Aspect Sentiment Analysis |
d8822680 | We explore six challenges for neural machine translation: domain mismatch, amount of training data, rare words, long sentences, word alignment, and beam search. We show both deficiencies and improvements over the quality of phrasebased statistical machine translation. | Six Challenges for Neural Machine Translation |
d236460112 | Understanding the multi-scale visual information in a video is essential for Video Question Answering (VideoQA). Therefore, we propose a novel Multi-Scale Progressive Attention Network (MSPAN) to achieve relational reasoning between cross-scale video information. We construct clips of different lengths to represent different scales of the video. Then, the cliplevel features are aggregated into node features by using max-pool, and a graph is generated for each scale of clips. For cross-scale feature interaction, we design a message passing strategy between adjacent scale graphs, i.e., topdown scale interaction and bottom-up scale interaction. Under the question's guidance of progressive attention, we realize the fusion of all-scale video features. Experimental evaluations on three benchmarks: TGIF-QA, MSVD-QA and MSRVTT-QA show our method has achieved state-of-the-art performance. | Multi-Scale Progressive Attention Network for Video Question Answering |
d15532760 | The TIPSTER Text Summarization Evaluation (SUMMAC) has established definitively that automatic text summarization is very effective in relevance assessment tasks. Summaries as short as 17% of full text length sped up decisionmaking by almost a factor of 2 with no statistically significant degradation in Fscore accuracy. SUMMAC has also introduced a new intrinsic method for automated evaluation of informative summaries. | The TIPSTER SUMMAC Text Summarization Evaluation |
d257663514 | ChatGPT, the first large language model with mass adoption, has demonstrated remarkable performance in numerous natural language tasks. Despite its evident usefulness, evaluating ChatGPT's performance in diverse problem domains remains challenging due to the closed nature of the model and its continuous updates via Reinforcement Learning from Human Feedback (RLHF). We highlight the issue of data contamination in ChatGPT evaluations, with a case study in stance detection. We discuss the challenge of preventing data contamination and ensuring fair model evaluation in the age of closed and continuously trained models. | Can we trust the evaluation on ChatGPT? |
d245144567 | We consider two problems of NMT domain adaptation using meta-learning. First, we want to reach domain robustness, i.e., we want to reach high quality on both domains seen in the training data and unseen domains. Second, we want our systems to be adaptive, i.e., making it possible to finetune systems with just hundreds of in-domain parallel sentences. We study the domain adaptability of meta-learning when improving the domain robustness of the model. In this paper, we propose a novel approach, RML-NMT (Robust Meta-Learning Framework for Neural Machine Translation Domain Adaptation), which improves the robustness of existing meta-learning models. More specifically, we show how to use a domain classifier in curriculum learning and we integrate the word-level domain mixing model into the metalearning framework with a balanced sampling strategy. Experiments on English→German and English→Chinese translation show that RMLNMT improves in terms of both domain robustness and domain adaptability in seen and unseen domains 1 . | Improving Both Domain Robustness and Domain Adaptability in Machine Translation |
d252624756 | We present the requirements, design guidelines, and the software architecture of an open-source toolkit dedicated to the pre-processing of sign language video material. The toolkit is a collection of functions and command-line tools designed to be integrated with build automation systems. Every pre-processing tool is dedicated to standard pre-processing operations (e.g., trimming, cropping, resizing) or feature extraction (e.g., identification of areas of interest, landmark detection) and can be used also as a standalone Python module. The UML diagrams of its architecture are presented together with a few working examples of its usage. The software is freely available with an open-source license on a public repository. | A Software Toolkit for Pre-processing Sign Language Video Streams |
d3059017 | It has remained an open question whether the twins property for weighted tree automata is decidable. This property is crucial for determinizing such an automaton, and it has been argued that determinization improves the output of parsers and translation systems. We show that the twins property for weighted tree automata over extremal semifields is decidable. | Deciding the Twins Property for Weighted Tree Automata over Extremal Semifields |
d249062526 | Data sparsity is a main problem hindering the development of code-switching (CS) NLP systems. In this paper, we investigate data augmentation techniques for synthesizing dialectal Arabic-English CS text. We perform lexical replacements using word-aligned parallel corpora where CS points are either randomly chosen or learnt using a sequence-to-sequence model. We compare these approaches against dictionary-based replacements. We assess the quality of the generated sentences through human evaluation and evaluate the effectiveness of data augmentation on machine translation (MT), automatic speech recognition (ASR), and speech translation (ST) tasks. Results show that using a predictive model results in more natural CS sentences compared to the random approach, as reported in human judgements. In the downstream tasks, despite the random approach generating more data, both approaches perform equally (outperforming dictionary-based replacements). Overall, data augmentation achieves 34% improvement in perplexity, 5.2% relative improvement on WER for ASR task, +4.0-5.1 BLEU points on MT task, and +2.1-2.2 BLEU points on ST over a baseline trained on available data without augmentation. | Investigating Lexical Replacements for Arabic-English Code-Switched Data Augmentation |
d12331876 | To extract structured representations of newsworthy events from Twitter, unsupervised models typically assume that tweets involving the same named entities and expressed using similar words are likely to belong to the same event. Hence, they group tweets into clusters based on the cooccurrence patterns of named entities and topical keywords. However, there are two main limitations. First, they require the number of events to be known beforehand, which is not realistic in practical applications. Second, they don't recognise that the same named entity might be referred to by multiple mentions and tweets using different mentions would be wrongly assigned to different events. To overcome these limitations, we propose a nonparametric Bayesian mixture model with word embeddings for event extraction, in which the number of events can be inferred automatically and the issue of lexical variations for the same named entity can be dealt with properly. Our model has been evaluated on three datasets with sizes ranging between 2,499 and over 60 million tweets. Experimental results show that our model outperforms the baseline approach on all datasets by 5-8% in F-measure. | Event extraction from Twitter using Non-Parametric Bayesian Mixture Model with Word Embeddings |
d259138479 | Coherence is an important aspect of text quality, and various approaches have been applied to coherence modeling. However, existing methods solely focus on a single document's coherence patterns, ignoring the underlying correlation between documents. We investigate a GCN-based coherence model that is capable of capturing structural similarities between documents. Our model first creates a graph structure for each document, from where we mine different subgraph patterns. We then construct a heterogeneous graph for the training corpus, connecting documents based on their shared subgraphs. Finally, a GCN is applied to the heterogeneous graph to model the connectivity relationships. We evaluate our method on two tasks, assessing discourse coherence and automated essay scoring. Results show that our GCN-based model outperforms all baselines, achieving a new state-of-the-art on both tasks. | Modeling Structural Similarities between Documents for Coherence Assessment with Graph Convolutional Networks |
d251698581 | The study presented in this paper demonstrates how transcribed podcast material di ers with respect to lexical content from other collections of English language data: editorial text, social media, both long form and microblogs, dialogue from movie scripts, and transcribed phone conversations. Most of the recorded di erences are as might be expected, reflecting known or assumed di erence between spoken and wri en language, between dialogue and soliloquy, and between scripted formal and unscripted informal language use. Most notably, podcast material, compared to the hitherto typical training sets from editorial media, is characterised by being in the present tense, and with a much higher incidence of pronouns, interjections, and negations. These characteristics are, unsurprisingly, largely shared with social media texts. Where podcast material di ers from social media material is in its a itudinal content, with many more amplifiers and much less negative a itude than in blog texts. This variation, besides being of philological interest, has ramifications for computational work. Information access for material which is not primarily topical should be designed to be sensitive to such variation that defines the data set itself and discriminates items within it. In general, training sets for language models are a non-trivial parameter which are likely to show e ects both expected and unexpected when applied to data from other sources and the characteristics and provenance of data used to train a model should be listed on the label as a minimal form of downstream consumer protection. | Lexical variation in English language podcasts, editorial media, and social media |
d256826883 | We use insights from research on American Sign Language (ASL) phonology to train models for isolated sign language recognition (ISLR), a step towards automatic sign language understanding. Our key insight is to explicitly recognize the role of phonology in sign production to achieve more accurate ISLR than existing work which does not consider sign language phonology. We train ISLR models that take in pose estimations of a signer producing a single sign to predict not only the sign but additionally its phonological characteristics, such as the handshape. These auxiliary predictions lead to a nearly 9% absolute gain in sign recognition accuracy on the WLASL benchmark, with consistent improvements in ISLR regardless of the underlying prediction model architecture. This work has the potential to accelerate linguistic research in the domain of signed languages and reduce communication barriers between deaf and hearing people. | Improving Sign Recognition with Phonology |
d234482105 | The adoption of natural language generation (NLG) models can leave individuals vulnerable to the generation of harmful information memorized by the models, such as conspiracy theories. While previous studies examine conspiracy theories in the context of social media, they have not evaluated their presence in the new space of generative language models. In this work, we investigate the capability of language models to generate conspiracy theory text. Specifically, we aim to answer: can we test pretrained generative language models for the memorization and elicitation of conspiracy theories without access to the model's training data? We highlight the difficulties of this task and discuss it in the context of memorization, generalization, and hallucination. Utilizing a new dataset consisting of conspiracy theory topics and machine-generated conspiracy theories helps us discover that many conspiracy theories are deeply rooted in the pretrained language models. Our experiments demonstrate a relationship between model parameters such as size and temperature and their propensity to generate conspiracy theory text. These results indicate the need for a more thorough review of NLG applications before release and an indepth discussion of the drawbacks of memorization in generative language models. Tom Brown, Dawn Song, Ulfar Erlingsson, et al. 2020. Extracting training data from large language models. arXiv preprint arXiv:2012.07805. | Investigating Memorization of Conspiracy Theories in Text Generation Note: This paper contains examples of potentially offensive conspiracy theory text |
d5850321 | A central component ofKallmeyer and Joshi 2003is the idea that the contribution of a quantifier is separated into a scope and a predicate argument part. Quantified NPs are analyzed as multi-component TAGs, where the scope part of the quantifier introduces the proposition containing the quantifier, and the predicate-argument part introduces the restrictive clause. This paper shows that this assumption presents difficulties for the compositional interpretation of NP coordination structures, and proposes an analysis which is based on LTAG semantics with semantic unification, developed in Kallmeyer and Romero 2004. | LTAG Semantics of NP-Coordination * |
d245856801 | This paper presents a comparative evaluation of four commercial ASR systems which are evaluated according to the post-editing effort required to reach "publishable" quality and according to the number of errors they produce. For the error annotation task, an original error typology for transcription errors is proposed. This study also seeks to examine whether there is a difference in the performance of these systems between native and non-native English speakers. The experimental results suggest that among the four systems, Trint and Microsoft obtain the best scores. It is also observed that most systems perform noticeably better with native speakers and that all systems are most prone to fluency errors. | Translation and Interpreting Technology Online |
d252624444 | Throughout the COVID-19 pandemic, a parallel infodemic has also been going on such that the information has been spreading faster than the virus itself. During this time, every individual needs to access accurate news in order to take corresponding protective measures, regardless of their country of origin or the language they speak, as misinformation can cause significant loss to not only individuals but also society. In this paper we train several machine learning models (ranging from traditional machine learning to deep learning) to try to determine whether news articles come from either a reliable or an unreliable source, using just the body of the article. Moreover, we use a previously introduced corpus of news in Swedish related to the COVID-19 pandemic for the classification task. Given that our dataset is both unbalanced and small, we use subsampling and easy data augmentation (EDA) to try to solve these issues. In the end, we realize that, due to the small size of our dataset, using traditional machine learning along with data augmentation yields results that rival those of transformer models such as BERT. | A First Attempt at Unreliable News Detection in Swedish |
d240288892 | Mental health is a critical issue in modern society, and mental disorders could sometimes turn to suicidal ideation without adequate treatment. Early detection of mental disorders and suicidal ideation from social content provides a potential way for effective social intervention. Recent advances in pretrained contextualized language representations have promoted the development of several domain-specific pretrained models and facilitated several downstream applications. However, there are no existing pretrained language models for mental healthcare. This paper trains and releases two pretrained masked language models, i.e., MentalBERT and MentalRoBERTa, to benefit machine learning for the mental healthcare research community. Besides, we evaluate our trained domain-specific models and several variants of pretrained language models on several mental disorder detection benchmarks and demonstrate that language representations pretrained in the target domain improve the performance of mental health detection tasks. | MentalBERT: Publicly Available Pretrained Language Models for Mental Healthcare |
d102350767 | Data-to-text generation can be conceptually divided into two parts: ordering and structuring the information (planning), and generating fluent language describing the information (realization). Modern neural generation systems conflate these two steps into a single end-to-end differentiable system. We propose to split the generation process into a symbolic text-planning stage that is faithful to the input, followed by a neural generation stage that focuses only on realization. For training a plan-to-text generator, we present a method for matching reference texts to their corresponding text plans. For inference time, we describe a method for selecting high-quality text plans for new inputs. We implement and evaluate our approach on the WebNLG benchmark. Our results demonstrate that decoupling text planning from neural realization indeed improves the system's reliability and adequacy while maintaining fluent output. We observe improvements both in BLEU scores and in manual evaluations. Another benefit of our approach is the ability to output diverse realizations of the same input, paving the way to explicit control over the generated text structure. | Step-by-Step: Separating Planning from Realization in Neural Data-to-Text Generation * |
d202611495 | This paper presents the Instituto de Telecomunicações-Instituto Superior Técnico submission to Task 1 of the SIGMORPHON 2019 Shared Task. Our models combine sparse sequence-to-sequence models with a two-headed attention mechanism that learns separate attention distributions for the lemma and inflectional tags. Among submissions to Task 1, our models rank second and third. Despite the low data setting of the task (only 100 in-language training examples), they learn plausible inflection patterns and often concentrate all probability mass into a small set of hypotheses, making beam search exact. | IT-IST at the SIGMORPHON 2019 Shared Task: Sparse Two-headed Models for Inflection |
d251041146 | When writing, a person may need to anticipate questions from their audience, but different social groups may ask very different types of questions. If someone is writing about a problem they want to resolve, what kind of follow-up question will a domain expert ask, and could the writer better address the expert's information needs by rewriting their original post? In this paper, we explore the task of socially-aware question generation. We collect a data set of questions and posts from social media, including background information about the question-askers' social groups. We find that different social groups, such as experts and novices, consistently ask different types of questions. We train several text-generation models that incorporate social information, and we find that a discrete social-representation model outperforms the text-only model when different social groups ask highly different questions from one another. Our work provides a framework for developing text generation models that can help writers anticipate the information expectations of highly different social groups. | How Well Do You Know Your Audience? Toward Socially-aware Question Generation |
d51868339 | Neural machine translation (NMT) has significantly improved the quality of automatic translation models. One of the main challenges in current systems is the translation of rare words. We present a generic approach to address this weakness by having external models annotate the training data as Experts, and control the model-expert interaction with a pointer network and reinforcement learning. Our experiments using phrase-based models to simulate Experts to complement neural machine translation models show that the model can be trained to copy the annotations into the output consistently. We demonstrate the benefit of our proposed framework in outof-domain translation scenarios with only lexical resources, improving more than 1.0 BLEU point in both translation directions English→Spanish and German→English. | Towards one-shot learning for rare-word translation with external experts |
d202539655 | Recent efforts in cross-lingual word embedding (CLWE) learning have predominantly focused on fully unsupervised approaches that project monolingual embeddings into a shared cross-lingual space without any cross-lingual signal. The lack of any supervision makes such approaches conceptually attractive. Yet, their only core difference from (weakly) supervised projection-based CLWE methods is in the way they obtain a seed dictionary used to initialize an iterative self-learning procedure. The fully unsupervised methods have arguably become more robust, and their primary use case is CLWE induction for pairs of resourcepoor and distant languages. In this paper, we question the ability of even the most robust unsupervised CLWE approaches to induce meaningful CLWEs in these more challenging settings. A series of bilingual lexicon induction (BLI) experiments with 15 diverse languages (210 language pairs) show that fully unsupervised CLWE methods still fail for a large number of language pairs (e.g., they yield zero BLI performance for 87/210 pairs). Even when they succeed, they never surpass the performance of weakly supervised methods (seeded with 500-1,000 translation pairs) using the same self-learning procedure in any BLI setup, and the gaps are often substantial. These findings call for revisiting the main motivations behind fully unsupervised CLWE methods. | Do We Really Need Fully Unsupervised Cross-Lingual Embeddings? |
d45503863 | Unknown words are a challenge for any NLP task, including sentiment analysis. Here, we evaluate the extent to which sentiment polarity of complex words can be predicted based on their morphological make-up. We do this on German as it has very productive processes of derivation and compounding and many German hapax words, which are likely to bear sentiment, are morphologically complex. We present results of supervised classification experiments on new datasets with morphological parses and polarity annotations. | Evaluating the Morphological Compositionality of Polarity |
d257219356 | Conversational tutoring systems (CTSs) aim to help students master educational material with natural language interaction in the form of a dialog. CTSs have become a key pillar in educational data mining research. A key challenge in CTSs is to engage the student in the conversation while exposing them to a diverse set of teaching strategies, akin to a human teacher, thereby, helping them learn in the process. Different from previous work that generates responses given the strategies as input, we propose to jointly predict teaching strategies and generate tutor responses accordingly, which fits a more realistic application scenario. We benchmark several competitive models on three dialog tutoring datasets and propose a unified framework that combines teaching response generation and pedagogical strategy prediction, where a self-distillation mechanism is adopted to guide the teaching strategy learning and facilitate tutor response generation. Our experiments and analyses shed light on how teaching strategies affect dialog tutoring. | Strategize Before Teaching: A Conversational Tutoring System with Pedagogy Self-Distillation |
d252819344 | Document-level biomedical relation extraction (Bio-DocuRE) is an important branch of biomedical text mining that aims to automatically extract all relation facts from the biomedical text. Since there are a considerable number of relations in biomedical documents that need to be judged by other existing relations, logical reasoning has become a research hotspot in the past two years. However, current models with reasoning are single-granularity only based on one element information, ignoring the complementary fact of different granularity reasoning information. In addition, obtaining rich document information is a prerequisite for logical reasoning, but most of the previous models cannot sufficiently utilize document information, which limits the reasoning ability of the model. In this paper, we propose a novel Bio-DocuRE model called FILR, based on Multi-Dimensional Fusion Information and Multi-Granularity Logical Reasoning. Specifically, FILR presents a multi-dimensional information fusion module MDIF to extract sufficient global document information. Then FILR proposes a multi-granularity reasoning module MGLR to obtain rich inference information through the reasoning of both entity-pairs and mention-pairs. We evaluate our FILR model on two widely used biomedical corpora CDR and GDA 1 . Experimental results show that FILR achieves state-of-the-art performance. | Document-level Biomedical Relation Extraction Based on Multi-Dimensional Fusion Information and Multi-Granularity Logical Reasoning |
d18987584 | This paper describes the aspect extraction system submitted by IHS R&D Belarus team at the SemEval-2014 shared task related to Aspect-Based Sentiment Analysis. Our system is based on IHS Goldfire linguistic processor and uses a rich set of lexical, syntactic and statistical features in CRF model. We participated in two domain-specific tasksrestaurants and laptopswith the same system trained on a mixed corpus of reviews. Among submissions of constrained systems from 28 teams, our submission was ranked first in laptop domain and fourth in restaurant domain for the subtask A devoted to aspect extraction. | IHS R&D Belarus: Cross-domain Extraction of Product Features using Conditional Random Fields |
d201698235 | Gang-involved youth in cities such as Chicago sometimes post on social media to express their aggression towards rival gangs and previous research has demonstrated that a deep learning approach can predict aggression and loss in posts. To address the possibility of bias in this sensitive application, we developed an approach to systematically interpret the state of the art model. We found, surprisingly, that it frequently bases its predictions on stop words such as "a" or "on", an approach that could harm social media users who have no aggressive intentions. To tackle this bias, domain experts annotated the rationales, highlighting words that explain why a tweet is labeled as "aggression". These new annotations enable us to quantitatively measure how justified the model predictions are, and build models that drastically reduce bias. Our study shows that in high stake scenarios, accuracy alone cannot guarantee a good system and we need new evaluation methods. | Detecting and Reducing Bias in a High Stakes Domain |
d516338 | We present the CLiPS Stylometry Investigation (CSI) corpus, a new Dutch corpus containing reviews and essays written by university students. It is designed to serve multiple purposes: detection of age, gender, authorship, personality, sentiment, deception, topic and genre. Another major advantage is its planned yearly expansion with each year's new students. The corpus currently contains about 305,000 tokens spread over 749 documents. The average review length is 128 tokens; the average essay length is 1126 tokens. The corpus will be made available on the CLiPS website (www.clips.uantwerpen.be/datasets) and can freely be used for academic research purposes. An initial deception detection experiment was performed on this data. Deception detection is the task of automatically classifying a text as being either truthful or deceptive, in our case by examining the writing style of the author. This task has never been investigated for Dutch before. We performed a supervised machine learning experiment using the SVM algorithm in a 10-fold cross-validation setup. The only features were the token unigrams present in the training data. Using this simple method, we reached a state-of-the-art F-score of 72.2%. | CLiPS Stylometry Investigation (CSI) corpus: A Dutch corpus for the detection of age, gender, personality, sentiment and deception in text |
d258841031 | Extracting structured and grounded fact triples from raw text is a fundamental task in Information Extraction (IE). Existing IE datasets are typically collected from Wikipedia articles, using hyperlinks to link entities to the Wikidata knowledge base. However, models trained only on Wikipedia have limitations when applied to web domains, which often contain noisy text or text that does not have any factual information. We present WEBIE, the first large-scale, entity-linked closed IE dataset consisting of 1.6M sentences automatically collected from the English Common Crawl corpus. WEBIE also includes negative examples, i.e. sentences without fact triples, to better reflect the data on the web. We annotate ∼21K triples from WEBIE through crowdsourcing and introduce mWEBIE, a translation of the annotated set in four other languages: French, Spanish, Portuguese, and Hindi. We evaluate the in-domain, out-of-domain, and zero-shot cross-lingual performance of generative IE models and find models trained on WEBIE show better generalisability. We also propose three training strategies that use entity linking as an auxiliary task. Our experiments show that adding Entity-Linking objectives improves the faithfulness of our generative IE models 1 . | WEBIE: Faithful and Robust Information Extraction on the Web |
d202769374 | Similarity measures based purely on word embeddings are comfortably competing with much more sophisticated deep learning and expert-engineered systems on unsupervised semantic textual similarity (STS) tasks. In contrast to commonly used geometric approaches, we treat a single word embedding as e.g. 300 observations from a scalar random variable. Using this paradigm, we first illustrate that similarities derived from elementary pooling operations and classic correlation coefficients yield excellent results on standard STS benchmarks, outperforming many recently proposed methods while being much faster and trivial to implement. Next, we demonstrate how to avoid pooling operations altogether and compare sets of word embeddings directly via correlation operators between reproducing kernel Hilbert spaces. Just like cosine similarity is used to compare individual word vectors, we introduce a novel application of the centered kernel alignment (CKA) as a natural generalisation of squared cosine similarity for sets of word vectors. Likewise, CKA is very easy to implement and enjoys very strong empirical results. | Correlations between Word Vector Sets |
d202771243 | We present a method for extracting causality knowledge from Wikipedia, such as Protectionism → Trade war, where the cause and effect entities correspond to Wikipedia articles. Such causality knowledge is easy to verify by reading corresponding Wikipedia articles, to translate to multiple languages through Wikidata, and to connect to knowledge bases derived from Wikipedia. Our method exploits Wikipedia article sections that describe causality and the redundancy stemming from the multilinguality of Wikipedia. Experiments showed that our method achieved precision and recall above 98% and 64%, respectively. In particular, it could extract causalities whose cause and effect were written distantly in a | Weakly Supervised Multilingual Causality Extraction from Wikipedia |
d202774579 | Word embeddings have been widely used to study gender stereotypes in texts. One key problem regarding existing bias scores is to evaluate their validities: do they really reflect true bias levels? For a small set of words (e.g. occupations), we can rely on human annotations or external data. However, for most words, evaluating the correctness of them is still an open problem. In this work, we utilize word association test, which contains rich types of word connections annotated by human participants, to explore how gender stereotypes spread within our minds. Specifically, we use random walk on word association graph to derive bias scores for a large amount of words. Experiments show that these bias scores correlate well with bias in the real world. More importantly, comparing with word-embedding-based bias scores, it provides a different perspective on gender stereotypes in words. 1 | Exploring Human Gender Stereotypes with Word Association Test |
d202787714 | Solving long-lasting problems such as food insecurity requires a comprehensive understanding of interventions applied by governments and international humanitarian assistance organizations, and their results and consequences. Towards achieving this grand goal, a crucial first step is to extract past interventions and when and where they have been applied, from hundreds of thousands of reports automatically. In this paper, we developed a corpus annotated with interventions to foster research, and developed an information extraction system for extracting interventions and their location and time from text. We demonstrate early, very encouraging results on extracting interventions. | Towards Machine Reading for Interventions from Humanitarian-Assistance Program Literature |
d46939026 | We present the task of second language acquisition (SLA) modeling. Given a history of errors made by learners of a second language, the task is to predict errors that they are likely to make at arbitrary points in the future. We describe a large corpus of more than 7M words produced by more than 6k learners of English, Spanish, and French using Duolingo, a popular online language-learning app. Then we report on the results of a shared task challenge aimed studying the SLA task via this corpus, which attracted 15 teams and synthesized work from various fields including cognitive science, linguistics, and machine learning. | Second Language Acquisition Modeling |
d52936990 | Reading comprehension tasks test the ability of models to process long-term context and remember salient information. Recent work has shown that relatively simple neural methods such as the Attention Sum-Reader can perform well on these tasks; however, these systems still significantly trail human performance. Analysis suggests that many of the remaining hard instances are related to the inability to track entity-references throughout documents. This work focuses on these hard entity tracking cases with two extensions: (1) additional entity features, and (2) training with a multi-task tracking objective. We show that these simple modifications improve performance both independently and in combination, and we outperform the previous state of the art on the LAMBADA dataset, particularly on difficult entity examples. | Entity Tracking Improves Cloze-style Reading Comprehension |
d258463944 | This paper presents a study examining the role of classification overlap in second language (L2) vowel perception. Eighteen experienced L1 Mandarin Chinese (MC) listeners completed a classification task on twelve Australian English (AusE) monophthongal vowels. It was found that the AusE vowels were not equally difficult for the MC listeners, and the vowels also showed various overlapping patterns in the classification space. Correlation tests revealed that for individual L2 vowels, the classification accuracy was negatively correlated with the mean reaction time (RT), r = -0.817, p = 0.001; the level of classification overlap was negatively correlated with the mean accuracy, r = -0.860, p < 0.001, while positively correlated with the mean RT, r = 0.760, p = 0.004. The findings suggest that classification overlap can inform how experienced L2 listeners' first language (L1) phonology influences vowel perception in a non-native language. | Perceptual Overlap in Classification of L2 Vowels: Australian English Vowels Perceived by Experienced Mandarin Listeners |
d236460174 | Transformer-based pre-trained language models like BERT, though powerful in many tasks, are expensive in both memory and computation, due to their large number of parameters. Previous works show that some parameters in these models can be pruned away without severe accuracy drop. However, these redundant features contribute to a comprehensive understanding of the training data and removing them weakens the model's representation ability. In this paper, we propose GhostBERT, which generates more features with very cheap operations from the remaining features. In this way, GhostBERT has similar memory and computational cost as the pruned model, but enjoys much larger representation power. The proposed ghost module can also be applied to unpruned BERT models to enhance their performance with negligible additional parameters and computation. Empirical results on the GLUE benchmark on three backbone models (i.e., BERT, RoBERTa and ELECTRA) verify the efficacy of our proposed method. | GhostBERT: Generate More Features with Cheap Operations for BERT |
d61205942 | Recognizing textual entailment (RTE) has been proposed as a task in computational linguistics under a successful series of annual evaluation campaigns started in 2005 with the Pascal RTE-1 shared task. RTE is defined as the capability of a system to recognize that the meaning of a portion of text (usually one or few sentences) entails the meaning of another portion of text. Subsequently, the task has also been extended to recognizing specific cases of non-entailment, as when the meaning of the first text contradicts the meaning of the second text. Although the study of entailment phenomena in natural language was addressed much earlier, the novelty of the RTE evaluation was to propose a simple text-to-text task to compare human and system judgments, making it possible to build data sets and to experiment with a variety of approaches. Two main reasons likely contributed to the success of the initiative: First, the possibility to address, for the first time, the complexity of entailment phenomena under a data-driven perspective; second, the text-to-text approach allows one to easily incorporate a textual entailment engine into applications (e.g., question answering, summarization, information extraction) as a core inferential component.Given the large interest in RTE, a book on the topic is very much welcome. The book satisfies at least two needs in the community. On the one hand, it was time to collect and organize a large amount of research and publications produced in the last decade or so. Under this perspective a particular merit of the book is to set a homogeneous and general descriptive framework under which different approaches can be reported and compared with a uniform conceptual schema. On the other hand, after a decade of research in the field, it is the right time to open the discussion both to the results that have been achieved and to future challenges in RTE. With this perspective the book targets not only people interested in textual entailment but people interested in semantic inference and semantic processing at large, intended as a core capability of natural language understanding. Researchers working in related areas such as textual similarity, causality, and temporal reasoning, just to mention a few, may benefit from this monograph.The book is organized into six chapters, covering RTE motivations, task definitions, approaches and data-driven methodologies, relevant case studies, knowledge acquisition for RTE and, finally, future research perspectives.Chapter 1 provides the necessary context on textual entailment and RTE. Even a reader not familiar with the field can find in this chapter motivations, definitions, many | Recognizing Textual Entailment: Models and Applications |
d1173003 | Mapping phrases between languages as translation of each other by using an intermediate language (pivot language) may generate translation pairs that are wrong. Since a word or a phrase has different meanings in different contexts, we should map source and target phrases in an intelligent way. We propose a pruning method based on the context vectors to remove those phrase pairs that connect to each other by a polysemous pivot phrase or by weak translations. We use context vectors to implicitly disambiguate the phrase senses and to recognize irrelevant phrase translation pairs. Using the proposed method a relative improvement of 2.8 percent in terms of BLEU score is achieved. | Using Context Vectors in Improving a Machine Translation System with Bridge Language |
d23678406 | In this paper we show that reporting a single performance score is insufficient to compare non-deterministic approaches. We demonstrate for common sequence tagging tasks that the seed value for the random number generator can result in statistically significant (p < 10 −4 ) differences for state-of-the-art systems. For two recent systems for NER, we observe an absolute difference of one percentage point F 1 -score depending on the selected seed value, making these systems perceived either as state-of-the-art or mediocre. Instead of publishing and reporting single performance scores, we propose to compare score distributions based on multiple executions. Based on the evaluation of 50.000 LSTMnetworks for five sequence tagging tasks, we present network architectures that produce both superior performance as well as are more stable with respect to the remaining hyperparameters. The full experimental results are published in (Reimers and Gurevych, 2017). 1 The implementation of our network is publicly available. 2 | Reporting Score Distributions Makes a Difference: Performance Study of LSTM-networks for Sequence Tagging |
d259376711 | The naïve approach for fine-tuning pretrained deep learning models on downstream tasks involves feeding them mini-batches of randomly sampled data. In this paper, we propose a more elaborate method for fine-tuning Pretrained Multilingual Transformers (PMTs) on multilingual data. Inspired by the success of curriculum learning approaches, we investigate the significance of fine-tuning PMTs on multilingual data in a sequential fashion language by language. Unlike the curriculum learning paradigm where the model is presented with increasingly complex examples, we do not adopt a notion of "easy" and "hard" samples. Instead, our experiments draw insight from psychological findings on how the human brain processes new information and the persistence of newly learned concepts. We perform our experiments on a challenging news-framing dataset that contains texts in six languages. Our proposed method outperforms the naïve approach by achieving improvements of 2.57% in terms of F1 score. Even when we supplement the naïve approach with recency fine-tuning, we still achieve an improvement of 1.34% with a 3.63% convergence speed-up. Moreover, we are the first to observe an interesting pattern in which deep learning models exhibit a human-like primacyrecency effect. | BERTastic at SemEval-2023 Task 3: Fine-Tuning Pretrained Multilingual Transformers -Does Order Matter? |
d259376737 | We experiment with XLM-Twitter and XLM-RoBERTa models to predict the intimacy scores in Tweets i.e. the extent to which a Tweet contains intimate content. We propose a Transformer-TabNet based multimodal architecture using text data and statistical features from the text, which performs better than the vanilla Transformer based model. We further experiment with Adversarial Weight Perturbation to make our models generalized and robust. The ensemble of four of our best models achieve an over-all Pearson Coefficient of 0.5893 on the test dataset. | ODA_SRIB at SemEval-2023 Task 9: A Multimodal Approach for Improved Intimacy Analysis |
d7725084 | In this paper, we propose a graph kernel based approach for the automated extraction of protein-protein interactions (PPI) from scientific literature. In contrast to earlier approaches to PPI extraction, the introduced alldependency-paths kernel has the capability to consider full, general dependency graphs. We evaluate the proposed method across five publicly available PPI corpora providing the most comprehensive evaluation done for a machine learning based PPI-extraction system. Our method is shown to achieve state-of-theart performance with respect to comparable evaluations, achieving 56.4 F-score and 84.8 AUC on the AImed corpus. Further, we identify several pitfalls that can make evaluations of PPI-extraction systems incomparable, or even invalid. These include incorrect crossvalidation strategies and problems related to comparing F-score results achieved on different evaluation resources. | A Graph Kernel for Protein-Protein Interaction Extraction |
d235097367 | This paper presents the results of the shared task of the 2nd Workshop on Automatic Simultaneous Translation (AutoSimTrans). The task includes two tracks, one for text-to-text translation and one for speech-to-text, requiring participants to build systems to translate from either the source text or speech into the target text. Different from traditional machine translation, the AutoSimTrans shared task evaluates not only translation quality but also latency. We propose a metric "Monotonic Optimal Sequence" (MOS) considering both quality and latency to rank the submissions. We also discuss some important open issues in simultaneous translation. | Findings of the Second Workshop on Automatic Simultaneous Translation |
d249455488 | Spoken Language Understanding (SLU) models in industry applications are usually trained offline on historic data, but have to perform well on incoming user requests after deployment. Since the application data is not available at training time, this is formally similar to the domain generalization problem, where domains correspond to different temporal segments of the data, and the goal is to build a model that performs well on unseen domains, e.g., upcoming data. In this paper, we explore different strategies for achieving good temporal generalization, including instance weighting, temporal fine-tuning, learning temporal features and building a temporally-invariant model. Our results on data of large-scale SLU systems show that temporal information can be leveraged to improve temporal generalization for SLU models. | Temporal Generalization for Spoken Language Understanding |
d216144650 | Massively multilingual models for neural machine translation (NMT) are theoretically attractive, but often underperform bilingual models and deliver poor zero-shot translations. In this paper, we explore ways to improve them. We argue that multilingual NMT requires stronger modeling capacity to support language pairs with varying typological characteristics, and overcome this bottleneck via language-specific components and deepening NMT architectures. We identify the off-target translation issue (i.e. translating into a wrong target language) as the major source of the inferior zero-shot performance, and propose random online backtranslation to enforce the translation of unseen training language pairs. Experiments on OPUS-100 (a novel multilingual dataset with 100 languages) show that our approach substantially narrows the performance gap with bilingual models in both oneto-many and many-to-many settings, and improves zero-shot performance by ∼10 BLEU, approaching conventional pivot-based methods. 1 | Improving Massively Multilingual Neural Machine Translation and Zero-Shot Translation |
d252968341 | We introduce a new method to improve existing multilingual sentence embeddings with Abstract Meaning Representation (AMR). Compared with the original textual input, AMR is a structured semantic representation that presents the core concepts and relations in a sentence explicitly and unambiguously. It also helps reduce surface variations across different expressions and languages. Unlike most prior work that only evaluates the ability to measure semantic similarity, we present a thorough evaluation of existing multilingual sentence embeddings and our improved versions, which include a collection of five transfer tasks in different downstream applications. Experiment results show that retrofitting multilingual sentence embeddings with AMR leads to better state-of-the-art performance on both semantic textual similarity and transfer tasks. Our codebase and evaluation scripts can be found at https://github.com/jcyk/ MSE-AMR. | Retrofitting Multilingual Sentence Embeddings with Abstract Meaning Representation * |
d155100252 | Humans use language to refer to entities in the external world. Motivated by this, in recent years several models that incorporate a bias towards learning entity representations have been proposed. Such entity-centric models have shown empirical success, but we still know little about why.In this paper we analyze the behavior of two recently proposed entity-centric models in a referential task, Entity Linking in Multi-party Dialogue (SemEval 2018 Task 4). We show that these models outperform the state of the art on this task, and that they do better on lower frequency entities than a counterpart model that is not entity-centric, with the same model size. We argue that making models entitycentric naturally fosters good architectural decisions. However, we also show that these models do not really build entity representations and that they make poor use of linguistic context. These negative results underscore the need for model analysis, to test whether the motivations for particular architectures are borne out in how models behave when deployed.1 Note the analogy with traditional models in formal linguistics like Discourse Representation Theory (Kamp and Reyle, 2013).2 Source code for our model, the training procedure and the new dataset is published on https://github.com/ amore-upf/analysis-entity-centric-nns.3 https://competitions.codalab.org/ competitions/17310. | What do Entity-Centric Models Learn? Insights from Entity Linking in Multi-Party Dialogue |
d256597842 | Acronym Disambiguation (AD) is crucial for natural language understanding on various sources, including biomedical reports, scientific papers, and search engine queries. However, existing acronym disambiguation benchmarks and tools are limited to specific domains, and the size of prior benchmarks is rather small. To accelerate the research on acronym disambiguation, we construct a new benchmark named GLADIS with three components: (1) a much larger acronym dictionary with 1.5M acronyms and 6.4M long forms;(2) a pre-training corpus with 160 million sentences; (3) three datasets that cover the general, scientific, and biomedical domains. We then pre-train a language model, AcroBERT, on our constructed corpus for general acronym disambiguation, and show the challenges and values of our new benchmark.Unstructured TextPileStructured TextGiB CorporaSentenceArtificial intelligence (AI) is intelligence demonstrated by machines.Machine Learning (ML) is a method of data analysis that automates analytical model building. | GLADIS: A General and Large Acronym Disambiguation Benchmark |
d222091022 | Textbook Question Answering is a complex task in the intersection of Machine Comprehension and Visual Question Answering that requires reasoning with multimodal information from text and diagrams. For the first time, this paper taps on the potential of transformer language models and bottom-up and top-down attention to tackle the language and visual understanding challenges this task entails. Rather than training a language-visual transformer from scratch we rely on pretrained transformers, fine-tuning and ensembling. We add bottom-up and top-down attention to identify regions of interest corresponding to diagram constituents and their relationships, improving the selection of relevant visual information for each question and answer options. Our system ISAAQ reports unprecedented success in all TQA question types, with accuracies of 81.36%, 71.11% and 55.12% on true/false, text-only and diagram multiple choice questions. ISAAQ also demonstrates its broad applicability, obtaining state-of-theart results in other demanding datasets.• For the first time, we apply transformers to | ISAAQ -Mastering Textbook Questions with Pre-trained Transformers and Bottom-Up and Top-Down Attention |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.