paper_id
stringlengths 10
10
⌀ | yes_no
bool 2
classes | paper_index
int64 0
519
⌀ | evidence
stringlengths 0
37.7k
| question
stringlengths 4
11.7k
| answer
stringlengths 1
26k
|
|---|---|---|---|---|---|
null | false
| null |
Chiune Sugihara (杉原 千畝, Sugihara Chiune, 1 January 1900 – 31 July 1986) was a Japanese diplomat who served as vice-consul for the Japanese Empire in Kaunas, Lithuania. During the Second World War, Sugihara helped thousands of Jews flee Europe by issuing transit visas to them so that they could travel through Japanese territory, risking his job and the lives of his family. The fleeing Jews were refugees from German-occupied Western Poland and Soviet-occupied Eastern Poland, as well as residents of Lithuania. In 1985, the State of Israel honored Sugihara as one of the Righteous Among the Nations for his actions. He is the only Japanese national to have been so honored. The year 2020 was "The Year Chiune Sugihara" in Lithuania. It has been estimated as many as 100,000 people alive today are the descendants of the recipients of Sugihara visas.
|
How did Chiune Sugihara save lives?
|
He issued transit visas to Jews fleeing Europe during World War II, so they could travel through Japanese territory.
|
null | false
| null |
Who wrote the Great Gatsby?
|
The Great Gatsby was written by F. Scott Fitzgerald
|
|
null | false
| null |
What is the origin of Hello ?
|
According to the Oxford English Dictionary, hello is an alteration of hallo, hollo, which came from Old High German "halâ, holâ, emphatic imperative of halôn, holôn to fetch, used especially in hailing a ferryman". It also connects the development of hello to the influence of an earlier form, holla, whose origin is in the French holà (roughly, 'whoa there!', from French là 'there')
|
|
1802.05574
| false
| null |
The sentences and their corresponding triples were then divided. Each task contained 10 sentences and all of their unique corresponding triples from a particular OIE systems. Half of the ten sentences were randomly selected from SCI and the other half were randomly selected from WIKI. Crowd workers were asked to mark whether a triple was correct, namely, did the triple reflect the consequence of the sentence. Examples of correct and incorrect triples were provided. Complete labelling instructions and the presentation of the HITS can be found with the dataset. All triples were labelled by at least 5 workers.
Crowd workers were asked to mark whether a triple was correct, namely, did the triple reflect the consequence of the sentence.
|
What is the role of crowd-sourcing?
|
The answers are shown as follows:
* Crowd workers were asked to mark whether a triple was correct, namely, did the triple reflect the consequence of the sentence.
|
null | false
| null |
What is Acyl-CoA?
|
Acyl-CoA is a group of coenzymes that metabolize fatty acids. Acyl-CoA's are susceptible to beta oxidation, forming, ultimately, acetyl-CoA. The acetyl-CoA enters the citric acid cycle, eventually forming several equivalents of ATP. In this way, fats are converted to ATP, the universal biochemical energy carrier.
|
|
null | false
| 51
|
Automated, or robotic, journalism aims at news generation from structured data sources, either as the final product or as a draft for subsequent post-editing. At present, automated journalism typically focuses on domains such as sports, finance and similar statistics-based reporting, where there is a commercial product potential due to the high volume of news, combined with the expectation of a relatively straightforward task.
News generation systems—especially those deployed in practice—tend to be based on intricate template filling, aiming to give the users the full control of the generated facts, while maintaining a reasonable variability of the resulting text. This comes at the price of having to develop the templates and specify their control logic, neither of which are tasks naturally fitting journalists' work. Further, this development needs to be repeated for every domain, as the templates are not easily transferred across domains. Examples of the template-based news generation systems for Finnish are Voitto by the Finnish Public Service Broadcasting Company (YLE) used for sports news generation, as well as Vaalibotti BIBREF0, a hybrid machine learning and template-based system used for election news.
BIBREF1 suggested a neural template generation, which jointly models latent templates and text generation. Such a system increases interpretability and controllability of the generation, however, recent sequence-to-sequence systems represent the state-of-the-art in data-to-text generation. BIBREF2
In this paper, we report on the development of a news generation system for the Finnish ice hockey news domain, based on sequence-to-sequence methods. In order to train such a system, we compile a corpus of news based on over 2000 game reports from the Finnish News Agency STT. While developing this corpus into a form suitable for training of end-to-end systems naturally requires manual effort, we argue that compiling and refining a set of text examples is a more natural way for journalists to interact with the system, in order for them to codify their knowledge and to adapt it for new domains.
Our aim is to generate reports that give an overview of a game based on information inferrable from the statistics. Such reports can be used either as a basis for further post-editing by a journalist imprinting own insights and background information, or even used directly as a news stream labelled as machine-generated.
In the following, we will introduce the news dataset and the process of its creation, introduce an end-to-end model for news generation, and evaluate its output respective to the abovementioned objectives.
In order to train such a system, we compile a corpus of news based on over 2000 game reports from the Finnish News Agency STT.
|
How to compile a corpus by the authors?
|
The authors compile a corpus of news based on over 2000 game reports from the Finnish News Agency STT.
|
null | false
| null |
The Basilica of Santa Maria della Sanità is a basilica church located over the Catacombs of San Gaudioso, on a Piazza near where Via Sanità meets Via Teresa degli Scalzi, in the Rione of the Sanità, in Naples, Italy. The church is also called San Vincenzo or San Vincenzo della Sanità, due to the cult of an icon of San Vincenzo Ferrer, also called locally O' Monacone (the big monk).
|
Given the reference text below, what is the official name of the church San Vincenzo?
|
The church called San Vincenzo is officially known as the Basilica of Santa Maria della Sanità, located over the Catacombs of San Gaudioso.
|
null | false
| 73
|
With the steady growth in the commercial websites and social media venues, the access to users' reviews have become easier. As the amount of data that can be mined for opinion increased, commercial companies' interests for sentiment analysis increased as well. Sentiment analysis is an important part of understanding user behavior and opinions on products, places, or services.
Sentiment analysis has long been studied by the research community, leading to several sentiment-related resources such as sentiment dictionaries that can be used as features for machine learning models BIBREF0 , BIBREF1 , BIBREF2 , BIBREF3 . These resources help increase sentiment analysis accuracies; however, they are highly dependent on language and require researchers to build such resources for every language to process.
Feature engineering is a large part of the model building phase for most sentiment analysis and emotion detection models BIBREF4 . Determining the correct set of features is a task that requires thorough investigation. Furthermore, these features are mostly language and dataset dependent making it even further challenging to build models for different languages. For example, the sentiment and emotion lexicons, as well as pre-trained word embeddings are not completely transferable to other languages which replicates the efforts for every language that users would like to build sentiment classification models on. For languages and tasks where the data is limited, extracting these features, building language models, training word embeddings, and creating lexicons are big challenges. In addition to the feature engineering effort, the machine learning models' parameters also need to be tuned separately for each language to get the optimal results.
In this paper, we take a different approach. We build a reusable sentiment analysis model that does not utilize any lexicons. Our goal is to evaluate how well a generic model can be used to mine opinion in different languages where data is more limited than the language where the generic model is trained on. To that end, we build a training set that contains reviews from different domains in English (e.g., movie reviews, product reviews) and train a recurrent neural network (RNN) model to predict polarity of those reviews. Then focusing on a domain, we make the model specialized in that domain by using the trained weights from the larger data and further training with data on a specific domain. To evaluate the reusability of the sentiment analysis model, we test with non-English datasets. We first translate the test set to English and use the pre-trained model to score polarity in the translated text. In this way, our proposed approach eliminates the need to train language-dependent models, use of sentiment lexicons and word embeddings for each language. Our experiments show that a generalizable sentiment analysis model can be utilized successfully to perform opinion mining for languages that do not have enough resources to train specific models.
The contributions of this study are; 1) a robust approach that utilizes machine translation to reuse a model trained on one language in other languages, 2) an RNN-based approach to eliminate feature extraction as well as resource requirements for sentiment analysis, and 3) a technique that statistically significantly outperforms baselines for multilingual sentiment analysis task when data is limited. To the best of our knowledge, this study is the first to apply a deep learning model to the multilingual sentiment analysis task.
We build a reusable sentiment analysis model that does not utilize any lexicons.
|
What kind of model do they build?
|
A reusable sentiment analysis model that does not utilize any lexicons.
|
1903.03467
| false
| null |
We highlight the problem of translating between languages with different morphological systems, in which the target translation must contain gender and number information that is not available in the source. We propose a method for injecting such information into a pre-trained NMT model in a black-box setting. We demonstrate the effectiveness of this method by showing an improvement of 2.3 BLEU in an English-to-Hebrew translation setting where the speaker and audience gender can be inferred. We also perform a fine-grained syntactic analysis that shows how our method enables to control the morphological realization of first and second-person pronouns, together with verbs and adjectives related to them. In future work we would like to explore automatic generation of the injected context, or the use of cross-sentence context to infer the injected information.
We also perform a fine-grained syntactic analysis that shows how our method enables to control the morphological realization of first and second-person pronouns, together with verbs and adjectives related to them.
|
What conclusions are drawn from the syntactic analysis?
|
The answers are shown as follows:
* our method enables to control the morphological realization of first and second-person pronouns, together with verbs and adjectives related to them
|
null | false
| null |
Łęgonice [wɛnɡɔˈnit͡sɛ] is a village in the administrative district of Gmina Nowe Miasto nad Pilicą, within Grójec County, Masovian Voivodeship, in east-central Poland. It lies approximately 3 kilometres (2 mi) west of Nowe Miasto nad Pilicą, 36 km (22 mi) south-west of Grójec, and 74 km (46 mi) south-west of Warsaw.
The village has a population of 440.
|
What are close cities to Legonice?
|
Close cities to Legonice are Nowe Miasto nad Pilica, Grojec, and Warsaw.
|
1908.04531
| false
| null |
We base our annotation procedure on the guidelines and schemas presented in BIBREF0 , discussed in detail in section "Classification Structure" . As a warm-up procedure, the first 100 posts were annotated by two annotators (the author and the supervisor) and the results compared. This was used as an opportunity to refine the mutual understanding of the task at hand and to discuss the mismatches in these annotations for each sub-task.
In light of these findings our internal guidelines were refined so that no post should be labeled as offensive by interpreting any context that is not directly visible in the post itself and that any post containing any form of profanity should automatically be labeled as offensive. These stricter guidelines made the annotation procedure considerably easier while ensuring consistency. The remainder of the annotation task was performed by the author, resulting in 3600 annotated samples.
As a warm-up procedure, the first 100 posts were annotated by two annotators (the author and the supervisor) and the results compared.
The remainder of the annotation task was performed by the author, resulting in 3600 annotated samples.
|
Who were the annotators?
|
The answers are shown as follows:
* the author and the supervisor
|
null | false
| 489
|
Settings. In our main experiments, we set: the cost of taking a CUR, GOAL, SUB, or DO action to be 0.01 (we will consider other settings subsequently), the cost of calling DONE to terminate the main goal (i.e. task error) equal the (unweighted) length of the shortest-path from the agent's location to the goal, and the goal stack's size (L) to be 2. We compare our RL-learned interaction policy with a rule-based baseline that first takes the GOAL action and then randomly selects actions.
In each episode, we enforce that the rule-based policy can take at most X a + y actions of type a, where y ∼ Bernoulli(X a − X a ) and X a is a constant. We tune each X a on the validation sets so that the rule-based policy has the same average count of each action as the RL-learned policy. To prevent early termination, we enforce that the rule-based policy cannot take more DONE actions than SUB actions unless when its SUB action's budget is exhausted. We also construct a skyline where the interaction policy is also learned by RL but with an operation policy that executes subgoals perfectly. As discussed in §5, we are primarily interested in goal-finding success rate and will refer to this metric briefly as success rate.
Main Results (Table). To inspect the potential benefits of asking for additional information, we compute how much the operation policy π improves when it is supplied with dense information about the current and/or goal states. As seen, success rate of the operation policy is lifted dramatically when both the current-state and goal descriptions are dense (∼2× increase on UNSEENSTR, ∼5× on UNSEENOBJ, and ∼3× on UNSEENENV). We find that dense information about the current state is more helpful on UNSEENSTR, while dense information about the goal is more valuable on UNSEENOBJ and UNSEENENV. This is reasonable because on UNSEENSTR, the agent has been trained to find similar goals. In contrast, the initial goal descriptions in UNSEENOBJ and UNSEE-NENV are completely new to the agent, thus gathering more information about them is necessary.
Aided by our RL-learned interaction policy, the agent observes a substantial ∼2× increase in success rate on UNSEENSTR, ∼5× on UNSEENOBJ, and ∼7× on UNSEENENV, compared to when performing tasks using only the operation policy. In unseen environments, with its capability of requesting subgoals, the agent impressively doubles the success rate of the operation policy that has access to dense descriptions. The RL-learned interaction policy is significantly more effective than (a) How frequently does the interaction policy execute different actions, on average across different types of tasks? Subgoals are requested much more in unseen environments.
q q q q q 0.0 0.5 1.0 1.5 2.0 0.2 0.4 0.6 0.8 1.0
Fraction of episode Average actions q CUR GOAL SUB (b) Over the course of a trajectory, how does the frequency of different types of actions change in unseen environments? Subgoals are requested in the middle, goal information at the beginning.
Figure: Analyzing the behavior of the RLlearned interaction policy (on validation environments).
q q q q q q q q q 0 25 50 75 −3 −2 −1 log10(cost) Success rate (%) q unseen_str unseen_obj unseen_env (a) Effect of cost on success rate.
q q q q q q q q q 0 3 6 the rule-based baseline (+7.1% on UNSEENENV). Compared to a policy that calls GOAL at the beginning and calls CUR at every step (which is equivalent to the dense-d s -and-d g baseline), our policy achieves higher success rate on UNSEENENV while making two times fewer requests. This is due to the ability to request subgoals.
On UNSEENSTR and UNSEENOBJ, the RL-learned policy has not closed the gap with the operation policy that performs tasks with dense descriptions. Our investigation finds that limited information often causes the policy to not request information about the current state (i.e. taking CUR) and terminate prematurely or go to a wrong place. Encoding uncertainty in the current-state description (e.g.,;) is a plausible future direction for tackling this issue. Finally, results obtained by replacing the learned operation policy with one that behaves optimally on subgoals shows that further improving performance of the operation policy on shortdistance goals would effectively enhance the agent's performance on long-distance goals.
Behavior of the RL-Learned Interaction Policy. Figure characterizes behaviors of the RLlearned interaction policy in three evaluation conditions. We expect that tasks in UNSEENSTR are the easiest and those in UNSEENENV are the hardest. As the difficulty of the evaluation condition increases, the interaction policy issues more CUR, SUB, and DO actions. The average number of GOAL actions does not vary, showing that the interaction policy has correctly learned that making more than one goal-clarifying request is unnecessary. Figure illustrates the distribution of each action along the length of an episode in the validation UNSEENENV dataset. The GOAL action, if taken, is always taken only once and immediately in the first step. The number of CUR actions gradually decreases over time. The agent makes most SUB requests in the middle of an episode, after its has attempted but failed to accomplish the main goals. We observe similar patterns on the other two validation sets.
Effects of Varying Action Cost. As mentioned, we assign the same cost to each CUR, GOAL, SUB, or DO action. Figure demonstrates the effects of changing this cost on the success rate of the agent. Setting the cost equal to 0.5 makes it too costly to take any action, inducing a policy that always calls DONE in the first step and thus fails on all tasks. Overall, the success rate of the agent rises as we reduce the action cost. The increase in success rate is most visible in UNSEENENV and least visible in UNSEENSTR. Figure provides more insights. As the action cost decreases, we observe a growth in the number of SUB and DO actions taken by the interaction policy. Meanwhile, the numbers of CUR and GOAL actions are mostly static. Since requesting subgoals is more helpful in unseen environments than in seen environments, the increase in the number of SUB actions leads the more visible boost in success rate on UNSEENENV tasks.
Performing Tasks with Deeper Goal Stacks. In Table, we test the functionality of our framework with a stack size 3, allowing the agent to request subgoals of subgoals. As expected, success rate on UNSEENENV is boosted significantly (+11.9% compared to using a stack of size 2). Success rate on UNSEENOBJ is largely unchanged; we find that the agent makes more SUB requests (averagely 4.5 requests per episode compared to 1.0 request made when the stack size is 2), but doing focus on learning a teaching policy in addition to an advice-requesting policy. An important assumption in these papers is that the teacher must share a common action space with the learner. More recent frameworks relax this assumption by allowing the teacher to specify high-level subgoals instead of low-level actions. HARI can be viewed as a strict extension of these frameworks. It allows the human to specify not only subgoals, but also any additional information about the current state and the goal that agent can interpret. Moreover, HARI equips the agent with multiple communication intentions and teaches it to select the most useful intention to convey in a given situation. Another line of work employs standard RL communication protocol, where the human can only transfer knowledge through numerical scores or categorical feedback. propose a framework where the human advises the agent using a domain-specific language, specifying rules that can be incorporated into the agent's model. In contrast, HARI operates with a black-box agent model. extract features from various types of language feedback to construct a reward function for reinforcement learning. We instead focus on deployment-time communication and directly incorporate the human feedback as input to agent's operation policy.
Task-Oriented Dialog and Generating Natural Language Questions. HARI models a taskoriented dialog problem. Many variants of this problem requires the agent to compose specific questions. The dominant approach in these problems is to mimic pre-collected human utterances. As discussed previously, naively mirroring human external behavior cannot enable agents to understand the limits of their knowledge. We teach the agent to understand its intrinsic needs through interaction with the human and the environment rather than through imitation of human behaviors. Another related line of work concerns generating natural language explanations of model decisions.
In summary, this paper presents a general POMDP framework for modeling human-agent communication. While we demonstrate this framework on a simplified navigation problem, our framework can theoretically capture richer types of human-agent communication. Hence, an important empirical question is how well our formulation generalizes to richer environments with more complex interactions and state spaces. Enhancing the sample efficiency of the learning policy by exploiting the hierarchical policy structure is an exciting future direction. Furthermore, techniques for generating faithful explanations can be applied to enhance the specificity of the generated questions.
Compared to a policy that calls GOAL at the beginning and calls CUR at every step (which is equivalent to the dense-ds -and-dg baseline), our policy achieves higher success rate on UNSEENENV while making two times fewer requests. This is due to the ability to request subgoals.
|
Page 2: "The agent achieves this level of performance while issuing only 4.8 requests to the assistant on average, representing less than 1⁄4 of the total number of actions taken in a task execution." Why is this significant or relevant? The number of times the agent has to make requests is determined by how much information is missing in the evaluation task, compared to the task it was trained on -- which means that the exact numbers are arbitrary. Such references to arbitrary numbers happen in a few places through the paper.
|
¼ actions are requests: we are comparing our agent with the agent that asks all the time (the dense ds and dg skyline). Our agent outperforms this agent on unseen environments, while making fewer requests. We will revise this claim.
|
null | false
| null |
What are fast recipes to make for dinner?
|
Fast recipes for dinner include:
1. Tacos - corn tortillas, ground beef, diced tomatoes, onions and shredded cheese
2. Stir fry - cook rice or noodle with vegetables in a wok and use some teriyaki sauce
3. Breakfast for dinner - eggs are quick to cook and make nutritious dinner meals as well
|
|
1705.07368
| false
| null |
To design an interpretable word embedding model for small corpora, we identify novel connections between word embeddings and topic models, and adapt advances from topic modeling. Following the distributional hypothesis BIBREF23 , the skip-gram's word embeddings parameterize discrete probability distributions over words INLINEFORM0 which tend to co-occur, and tend to be semantically coherent – a property leveraged by the Gaussian LDA model of BIBREF21 . This suggests that these discrete distributions can be reinterpreted as topics INLINEFORM1 . We thus reinterpret the skip-gram as a parameterization of a certain supervised naive Bayes topic model (Table TABREF2 , top-right). In this topic model, input words INLINEFORM2 are fully observed “cluster assignments,” and the words in INLINEFORM3 's contexts are a “document.” The skip-gram differs from this supervised topic model only in the parameterization of the “topics” via word vectors which encode the distributions with a log-bilinear model. Note that although the skip-gram is discriminative, in the sense that it does not jointly model the input words INLINEFORM4 , we are here equivalently interpreting it as encoding a “conditionally generative” process for the context given the words, in order to develop probabilistic models that extend the skip-gram.
As in LDA, this model can be improved by replacing the naive Bayes assumption with a mixed membership assumption. By applying the mixed membership representation to this topic model version of the skip-gram, we obtain the model in the bottom-right of Table TABREF2 . After once again parameterizing this model with word embeddings, we obtain our final model, the mixed membership skip-gram (MMSG) (Table TABREF2 , bottom-left). In the model, each input word has a distribution over topics INLINEFORM0 . Each topic has a vector-space embedding INLINEFORM1 and each output word has a vector INLINEFORM2 (a parameter, not an embedding for INLINEFORM3 ). A topic INLINEFORM4 is drawn for each context, and the words in the context are drawn from the log-bilinear model using INLINEFORM5 : DISPLAYFORM0
To design an interpretable word embedding model for small corpora, we identify novel connections between word embeddings and topic models, and adapt advances from topic modeling. Following the distributional hypothesis BIBREF23 , the skip-gram's word embeddings parameterize discrete probability distributions over words INLINEFORM0 which tend to co-occur, and tend to be semantically coherent – a property leveraged by the Gaussian LDA model of BIBREF21
We thus reinterpret the skip-gram as a parameterization of a certain supervised naive Bayes topic model (Table TABREF2 , top-right).
As in LDA, this model can be improved by replacing the naive Bayes assumption with a mixed membership assumption. By applying the mixed membership representation to this topic model version of the skip-gram, we obtain the model in the bottom-right of Table TABREF2 . After once again parameterizing this model with word embeddings, we obtain our final model, the mixed membership skip-gram (MMSG)
|
Which techniques for word embeddings and topic models are used?
|
The answers are shown as follows:
* skip-gram
* LDA
|
1707.03764
| false
| null |
N-GrAM ranked first in all cases except for the language variety task. In this case, the baseline was the top-ranked system, and ours was second by a small margin. Our system significantly out-performed the baseline on the joint task, as the baseline scored significantly lower for the gender task than for the variety task.
Our system significantly out-performed the baseline on the joint task, as the baseline scored significantly lower for the gender task than for the variety task.
|
On which task does do model do best?
|
Variety prediction task
|
1805.12070
| false
| null |
Training: The baseline model was trained using RNNLM BIBREF25 . Then, we trained our LSTM models with different hidden sizes [200, 500]. All LSTMs have 2 layers and unrolled for 35 steps. The embedding size is equal to the LSTM hidden size. A dropout regularization BIBREF26 was applied to the word embedding vector and POS tag embedding vector, and to the recurrent output BIBREF27 with values between [0.2, 0.4]. We used a batch size of 20 in the training. EOS tag was used to separate every sentence. We chose Stochastic Gradient Descent and started with a learning rate of 20 and if there was no improvement during the evaluation, we reduced the learning rate by a factor of 0.75. The gradient was clipped to a maximum of 0.25. For the multi-task learning, we used different loss weights hyper-parameters INLINEFORM0 in the range of [0.25, 0.5, 0.75]. We tuned our model with the development set and we evaluated our best model using the test set, taking perplexity as the final evaluation metric. Where the latter was calculated by taking the exponential of the error in the negative log-form. INLINEFORM1
FLOAT SELECTED: Figure 1: Multi-Task Learning Framework
Then, we trained our LSTM models with different hidden sizes [200, 500].
FLOAT SELECTED: Figure 1: Multi-Task Learning Framework
|
What is the architecture of the model?
|
The answers are shown as follows:
* LSTM
|
null | false
| 148
|
Classical QA systems face two main challenges related to question analysis and answer extraction. Several QA approaches were proposed in the literature for the open domain BIBREF16 , BIBREF17 and the medical domain BIBREF18 , BIBREF19 , BIBREF20 . A variety of methods were developed for question analysis, focus (topic) recognition and question type identification BIBREF21 , BIBREF22 , BIBREF23 , BIBREF24 . Similarly, many different approaches tackled document or passage retrieval and answer selection and (re)ranking BIBREF25 , BIBREF26 , BIBREF27 .
An alternative approach consists in finding similar questions or FAQs that are already answered BIBREF28 , BIBREF29 . One of the earliest question answering systems based on finding similar questions and re-using the existing answers was FAQ FINDER BIBREF30 . Another system that complements the existing Q&A services of NetWellness is SimQ BIBREF2 , which allows retrieval of similar web-based consumer health questions. SimQ uses syntactic and semantic features to compute similarity between questions, and UMLS BIBREF31 as a standardized semantic knowledge source. The system achieves 72.2% precision, 78.0% recall and 75.0% F-score on NetWellness questions. However, the method was evaluated only on one question similarity dataset, and the retrieved answers were not evaluated.
The aim of the medical task at TREC 2017 LiveQA was to develop techniques for answering complex questions such as consumer health questions, as well as to identify relevant answer sources that can comply with the sensitivity of medical information retrieval.
The CMU-OAQA system BIBREF32 achieved the best performance of 0.637 average score on the medical task by using an attentional encoder-decoder model for paraphrase identification and answer ranking. The Quora question-similarity dataset was used for training. The PRNA system BIBREF33 achieved the second best performance in the medical task with 0.49 average score using Wikipedia as the first answer source and Yahoo and Google searches as secondary answer sources. Each medical question was decomposed into several subquestions. To extract the answer from the selected text passage, a bi-directional attention model trained on the SQUAD dataset was used.
Deep neural network models have been pushing the limits of performance achieved in QA related tasks using large training datasets. The results obtained by CMU-OAQA and PRNA showed that large open-domain datasets were beneficial for the medical domain. However, the best system (CMU-OAQA) relying on the same training data obtained a score of 1.139 on the LiveQA open-domain task.
While this gap in performance can be explained in part by the discrepancies between the medical test questions and the open-domain questions, it also highlights the need for larger medical datasets to support deep learning approaches in dealing with the linguistic complexity of consumer health questions and the challenge of finding correct and complete answers.
Another technique was used by ECNU-ICA team BIBREF34 based on learning question similarity via two long short-term memory (LSTM) networks applied to obtain the semantic representations of the questions. To construct a collection of similar question pairs, they searched community question answering sites such as Yahoo! and Answers.com. In contrast, the ECNU-ICA system achieved the best performance of 1.895 in the open-domain task but an average score of only 0.402 in the medical task. As the ECNU-ICA approach also relied on a neural network for question matching, this result shows that training attention-based decoder-encoder networks on the Quora dataset generalized better to the medical domain than training LSTMs on similar questions from Yahoo! and Answers.com.
The CMU-LiveMedQA team BIBREF20 designed a specific system for the medical task. Using only the provided training datasets and the assumption that each question contains only one focus, the CMU-LiveMedQA system obtained an average score of 0.353. They used a convolutional neural network (CNN) model to classify a question into a restricted set of 10 question types and crawled "relevant" online web pages to find the answers. However, the results were lower than those achieved by the systems relying on finding similar answered questions. These results support the relevance of similar question matching for the end-to-end QA task as a new way of approaching QA instead of the classical QA approaches based on Question Analysis and Answer Retrieval.
The aim of the medical task at TREC 2017 LiveQA was to develop techniques for answering complex questions such as consumer health questions, as well as to identify relevant answer sources that can comply with the sensitivity of medical information retrieval.
|
What is the aim of the medical task at TREC 2017 LiveQA?
|
The aim of the medical task at TREC 2017 LiveQA was to develop techniques for answering complex questions such as consumer health questions, as well as to identify relevant answer sources that can comply with the sensitivity of medical information retrieval.
|
null | false
| null |
Tell me whether the following people are athletes: LeBron James, Eileen Gu, Bjergsen, Rory McIlroy, Naomi Osaka.
|
All of these people are athletes with the exception of Bjergsen. Bjergsen is a League of Legends (LoL) player. While LoL is considered an e-sport, its players do not fit into the traditional definition of an athlete, which is generally defined as someone who competes in physical sports. Having said that, some would say that competitive e-sports require significant mental stamina and finger dexterity, so one could argue that Bjergsen is an athlete.
|
|
null | false
| null |
Identify which animal species is alive or extinct: Arctotherium, Eastern Lowland Gorilla
|
Eastern Lowland Gorilla is alive, Arctotherium is extinct.
|
|
null | false
| null |
The biggest earning markets to date have been Japan ($45.9 million), followed by France ($35.5 million), China ($32.8 million), the UK ($25.3 million), Brazil ($22.9 million), Australia ($19 million), Germany ($17 million), Italy ($15.9 million), and South Korea ($15.5 million).
|
Extract, in order from largest earning to smallest, the biggest earning markets for the premier of Moana, the 2016 Disney film.
|
Japan ($45.9 million), followed by France ($35.5 million), China ($32.8 million), the UK ($25.3 million), Brazil ($22.9 million), Australia ($19 million), Germany ($17 million), Italy ($15.9 million), and South Korea ($15.5 million).
|
null | false
| 506
|
In this section, we introduce our proposed CGPFL in detail. The key idea is to dynamically cluster the clients into K disjoint groups based on their uploaded local model updates, and then develop a generalized model for each group by aggregating the updates in each clusters. These generalized models are utilized to guide the training directions of personalized models and transfer fine-grained generalization to them. Both the personalized models and the group models are trained in parallel, so we can denote the model parameters in matrix form. The generalized models can be written as
In this paper, we use capital characters to represent matrices unless stated otherwise.
The generalized models can be written as ΩK = [ω1, . . . , ωk, . . . , ωK] ∈ Rd×K, and the corresponding local approximations are ΩI,R = [ω1,R, . . . , ωi,R, . . . , ωN,R], where R is the number of local iterations and ωi,R, ωk ∈ Rd, ∀i ∈ [N], k ∈ [K].
|
I am not sure if I understand ΩI and ΩK, defined at the beginning of Section 4. Could you explain them in more detail?
|
As explained in Question 2, to compute Ωk, k=1,2,...,K in a distributional manner, we utilize the local updates omegaj, j in Ck as the local estimations of omegak. Therefore, the elements in ΩI are the local estimations of the related generalized models in ΩK. In the revision paper, we use omegai,R |i=1,2,...,N and omegak|k=1,2,...,K to define ΩI,R and ΩK, respectively, where R is the number of local iterations.
|
null | false
| null |
Did David Benioff direct any episodes in season three of Game of Thrones?
|
David Benioff directed "Walk of Punishment" in season three of Game of Thrones
|
|
null | false
| 204
|
Ethical considerations are established as essential part in planning mental health research and most research projects undergo approval by an ethics committee. On the contrary, the computational linguistics community has started only recently to consider ethical questions BIBREF72 , BIBREF73 . Likely, this is because computational linguistics was traditionally concerned with publicly available, impersonal texts such as newspapers or texts published with some temporal distance, which left a distance between the text and author. Conversely, recent social media research often deals with highly personal information of living individuals, who can be directly affected by the outcomes BIBREF72 .
BIBREF72 discuss issues that can arise when constructing datasets from social media and conducting analyses or developing predictive models based on these data, which we review here in relation to our project: Demographic bias in sampling the data can lead to exclusion of minority groups, resulting in overgeneralisation of models based on these data. As discussed in the introduction, personal recovery research suffers from a bias towards English-speaking Western individuals of white ethnicity. By studying multilingual accounts of ethnically diverse populations we explicitly address the demographic bias of previous research. Topic overexposure is tricky to address, where certain groups are perceived as abnormal when research repeatedly finds that their language is different or more difficult to process. Unlike previous research BIBREF45 , BIBREF47 , BIBREF46 our goal is not to reveal particularities in the language of individuals affected by mental health problems. Instead, we will compare accounts of individuals with BD from different settings (structured interviews versus informal online discourse) and of different backgrounds. While the latter bears the risk to overexpose certain minority groups, we will pay special attention to this in the dissemination of our results.
Lastly, most research, even when conducted with the best intentions, suffers from the dual-use problem BIBREF74 , in that it can be misused or have consequences that affect people's life negatively. For this reason, we refrain from publishing mental health classification methods, which could be used, for example, by health insurance companies for the risk assessment of applicants based on their social media profiles.
If and how informed consent needs to be obtained for research on social media data is a debated issue BIBREF75 , BIBREF76 , BIBREF77 , mainly because it is not straightforward to determine if posts are made in a public or private context. From a legal point of view, the privacy policies of Twitter and Reddit, explicitly allow analysis of the user contents by third party, but it is unclear to what extent users are aware of this when posting to these platforms BIBREF78 . However, in practice it is often infeasible to seek retrospective consent from hundreds or thousands of social media users. According to current ethical guidelines for social media research BIBREF79 , BIBREF80 and practice in comparable research projects BIBREF81 , BIBREF78 , it is regarded as acceptable to waive explicit consent if the anonymity of the users is preserved. Therefore, we will not ask the account holders of Twitter and Reddit posts included in our datasets for their consent.
BIBREF79 formulate guidelines for ethical social media health research that pertain especially to data collection and sharing. In line with these, we will only share anonymised and paraphrased excerpts from the texts, as it is often possible to recover a user name via a web search for the verbatim text of a post. However, we will make the original texts available as datasets to subsequent research under a data usage agreement. Since the (automatic) annotation of demographic variables in parts of our dataset constitutes especially sensitive information on minority status in conjunction with mental health, we will only share these annotations with researchers that demonstrate a genuine need for them, i.e. to verify our results or to investigate certain research questions.
Another important question is in which situations of encountering content indicative of a risk of self-harm or harm to others it would be appropriate or even required by duty of care for the research team to pass on information to authorities. Surprisingly, we could only find two mentions of this issue in social media research BIBREF81 , BIBREF82 . Acknowledging that suicidal ideation fluctuates BIBREF83 , we accord with the ethical review board's requirement in BIBREF81 to only analyse content posted at least three months ago. If the research team, which includes clinical psychologists, still perceives users at risk we will make use of the reporting facilities of Twitter and Reddit.
As a central component we consider the involvement of individuals with lived experience in our project, an aspect which is missing in the discussion of ethical social media health research so far. The proposal has been presented to an advisory board of individuals with a BD diagnosis and was received positively. The advisory board will be consulted at several stages of the project to inform the research design, analysis, and publication of results. We believe that board members can help to address several of the raised ethical problems, e.g., shaping the research questions to avoid feeding into existing biases or overexposing certain groups and highlighting potentially harmful interpretations and uses of our results.
By studying multilingual accounts of ethnically diverse populations we explicitly address the demographic bias of previous research.
|
How do the authors address the demographic bias of previous research?
|
By studying multilingual accounts of ethnically diverse populations the authors explicitly address the demographic bias of previous research.
|
null | false
| null |
What are the top nominated films to the Academy Awards in 2023?
|
The top nominated films for the 2023 Academy Awards are as follows:
Everything Everywhere All at Once - 11 nominations
All Quiet on the Western Front - 9 nominations
The Banshees of Inisherin - 9 nominations
Elvis - 8 nominations
The Fabelmans - 7 nominations
Tár - 6 nominations
Top Gun: Maverick - 6 nominations
Black Panther: Wakanda Forever - 5 nominations
Avatar: The Way of Water - 4 nominations
Babylon - 3 nominations
The Batman - 3 nominations
Triangle of Sadness - 3 nominations
The Whale - 3 nominations
|
|
null | false
| null |
"Chengtu, is a sub-provincial city which serves as the capital of the Chinese province of Sichuan. With a population of 20,937,757 inhabitants during the 2020 Chinese census, it is the fourth most populous city in China, and it is the only city apart from the four direct-administered municipalities with a population of over 20 million. It is traditionally the hub in Southwest China.
Chengdu is located in central Sichuan. The surrounding Chengdu Plain is known as the ""Country of Heaven"" (Chinese: 天府之国; pinyin: Tiānfǔ zhi Guó) and the ""Land of Abundance"". Its prehistoric settlers included the Sanxingdui culture. The site of Dujiangyan, an ancient irrigation system, is designated as a World Heritage Site. The Jin River flows through the city. Chengdu's culture largely reflects that of its province, Sichuan; in 2011, it was recognized by UNESCO as a city of gastronomy. It is associated with the giant panda, a Chinese national symbol, which inhabits the area of Sichuan; the city is home to the Chengdu Research Base of Giant Panda Breeding."
|
Which animal is associated with Chengdu?
|
Chengdu is associated with the giant panda, a Chinese national symbol, which inhabits the area of Sichuan; the city is home to the Chengdu Research Base of Giant Panda Breeding.
|
null | false
| null |
Why we need music?
|
I think the reason why we need music is that it adds color to our lives and provides another way to express emotions. We need alternative ways to express our emotions, whether we're happy or sad. Music is also important for events such as parties, ceremonies, social activities etc...
|
|
null | false
| null |
What are some common proteins in tacos?
|
Tacos can be filled with many types of proteins, such as chicken, ground meat, carnitas (shredded pork), carne asada (steak), fish, shrimp, tofu, black beans, refried beans, or lentils.
|
|
null | false
| null |
What is a spiral staircase?
|
A spiral staircase is a set of stairs that circles around a central pole and allows for a steeper ascent than other stair designs.
|
|
null | false
| null |
Classify these programming languages to compiled or interpreted: shell script, erlang, rust, python, javascript, C++, go
|
- Compiled language: erlang, rust, C++, go
- Interpreted language: shell script, python, javascript
|
|
null | false
| 112
|
The paper studies the effectiveness of various affect lexicons word embeddings to estimate emotional intensity in tweets. A light-weight easy to use affect computing framework (EmoInt) to facilitate ease of experimenting with various lexicon features for text tasks is open-sourced. It provides plug and play access to various feature extractors and handy scripts for creating ensembles.
Few problems explained in the analysis section can be resolved with the help of sentence embeddings which take the context information into consideration. The features used in the system are generic enough to use them in other affective computing tasks on social media text, not just tweet data. Another interesting feature of lexicon-based systems is their good run-time performance during prediction, future work to benchmark the performance of the system can prove vital for deploying in a real-world setting.
A light-weight easy to use affect computing framework (EmoInt) to facilitate ease of experimenting with various lexicon features for text tasks is open-sourced. It provides plug and play access to various feature extractors and handy scripts for creating ensembles.
|
What are the conclusions of this study?
|
A light-weight easy to use affect computing framework (EmoInt) to facilitate ease of experimenting with various lexicon features for text tasks is open-sourced. It provides plug and play access to various feature extractors and handy scripts for creating ensembles.
|
null | false
| 231
|
In a preliminary task, we looked for words which may designate sentences associated with controversial concepts. To this end, we ranked the words appearing in positive sentences according to their information gain for this task. The top of the list comprises the following: that, sexual, people, movement, religious, issues, rights.
The Wikipedia list of controversial issues specifies categories for the listed concepts, like Politics and economics, Religion, History, and Sexuality (some concepts are associated with two or more categories). While some top-ranked words - that, people, issues - do seem to directly indicate controversiality BIBREF12, BIBREF13, others seem to have more to do with the category they belong to. Although these categories may indeed indicate controversiality, we consider this as an indirect or implicit indication, since it is more related to the controversial theme than to controversiality per-se.
To control for this effect, we performed a second experiment where we set the concepts from one category as the test set, and used the others for training (concepts associated with the excluded category are left out, regardless of whether they are also associated with one of the training categories). We did this for 5 categories: History, Politics and economics, Religion, Science, and Sexuality. This way, thematic relatedness observed in the training set should have little or no effect on correctly estimating the level of controversy associated of concepts in the test set, and may even “mislead” the estimator. We note that previous work on controversiality does not seem to address this issue, probably because the meta-data used is less sensitive to it.
In a preliminary task, we looked for words which may designate sentences associated with controversial concepts. To this end, we ranked the words appearing in positive sentences according to their information gain for this task.
|
In a preliminary task, they ranked the words appearing in positive sentences according to what?
|
According to their information gain for this task.
|
null | false
| null |
What does #956 mean?
|
#956 refers to the Rio Grande Valley, an area in South Texas.The number 956 is the area code. It includes the city of McAllen, Pharr, Brownsville, and the popular beach destination: South Padre Island, just to name a few. People also say "Puro 956" to express pride for the area. The area shares the border with Mexico which explains why the majority of the population has Mexican heritage and the #1 language is Spanish. The 956 area is known for its rich culture and delicious Mexican food.
|
|
null | false
| null |
On 13 September 2007, on the back of India's victory at the 2007 T20 World Cup, BCCI announced a franchise-based Twenty20 cricket competition called Indian Premier League. The first season was slated to start in April 2008, in a "high-profile ceremony" in New Delhi. BCCI vice-president Lalit Modi, who spearheaded the IPL effort, spelled out the details of the tournament including its format, the prize money, franchise revenue system and squad composition rules. It was also revealed that the IPL would be run by a seven-man governing council composed of former India players and BCCI officials and that the top two teams of the IPL would qualify for that year's Champions League Twenty20. Modi also clarified that they had been working on the idea for two years and that the IPL was not started as a "knee-jerk reaction" to the ICL. The league's format was similar to that of the Premier League of England and the NBA in the United States.
In order to decide the owners for the new league, an auction was held on 24 January 2008 with the total base prices of the franchises costing around $400 million. At the end of the auction, the winning bidders were announced, as well as the cities the teams would be based in: Bangalore, Chennai, Delhi, Hyderabad, Jaipur, Kolkata, Mohali, and Mumbai. In the end, the franchises were all sold for a total of $723.59 million. The Indian Cricket League soon folded in 2008.
|
Which cities participated in the first IPL?
|
Bangalore, Chennai, Delhi, Hyderabad, Jaipur, Kolkata, Mohali, Mumbai
|
1808.04314
| false
| null |
Morphology deals with the internal structure of words BIBREF0 , BIBREF1 . Languages of the world have different word production processes. Morphological richness vary from language to language, depending on their linguistic typology. In natural language processing (NLP), taking into account the morphological complexity inherent to each language could be important for improving or adapting the existing methods, since the amount of semantic and grammatical information encoded at the word level, may vary significantly from language to language.
Additionally, most of the previous works do not analyze how the complexity changes when different types of morphological normalization procedures are applied to a language, e.g., lemmatization, stemming, morphological segmentation. This information could be useful for linguistic analysis and for measuring the impact of different word form normalization tools depending of the language. In this work, we analyze how the type-token relationship changes using different types of morphological normalization techniques.
In natural language processing (NLP), taking into account the morphological complexity inherent to each language could be important for improving or adapting the existing methods, since the amount of semantic and grammatical information encoded at the word level, may vary significantly from language to language.
This information could be useful for linguistic analysis and for measuring the impact of different word form normalization tools depending of the language.
|
what is the practical application for this paper?
|
Improve existing NLP methods. Improve linguistic analysis. Measure impact of word normalization tools.
|
null | false
| null |
You have been placed in the position of supporting the 2010 Supreme Court decision in Citizens United v. Federal Election Commission. Please develop your position into a one-page position paper (12-point font, single spaced).
|
The Citizens United case is ultimately one pertaining to freedom of speech—while undeniable to US citizens, the protection of free speech was previously denied to unions and corporations. Now, they can express support or criticism of political candidates. This should be good news for both dominant political parties, as liberals tending to support unions and conservatives tending to laud free enterprise both have great voice in the election process. Concern, however, has been raised that companies now have too loud a voice.
But, who makes up companies? Individual humans, who are free to join and leave those companies at will. Why would a group of humans be devoid of any rights to which the individuals of the group are entitled? The constitutional right to free speech, it follows, applies to groups of individuals – which companies are – and those groups should be able to leverage their earned dollars to further promote their own well-being. And the ruling has legal precedent: dating back to 1868, US corporations have enjoyed equal protection under the 14th amendment and protection against search and seizure without warrant (Totenberg, N.)
But the decision cannot be analyzed solely through legal theory. What have the actual impacts been since the ruling? Rather than observing unlimited spending from corporations as feared, the last ten years have reemphasized the importance of low-level donors; instead of focusing power on just one party or candidate that certain corporations back, political parties have taken turns assuming power; corporate donations remain under 10% of total political spending at the federal level. Furthermore, it seems that wealthy individuals, rather than any specific union or corporation, have been making oversized donations that dominant the budgets of super-PACs. It follows that any concerns around voices dominating the political conversation are indeed individuals, not unions or corporations as fear.
Citizens United allows companies to further act in their best interests and raise issues pertaining to the business. US citizens can and should be trusted to consume rhetoric intellectually, instead of shielded from the arguments companies are making.
|
|
null | false
| null |
In computability theory, the halting problem is the problem of determining, from a description of an arbitrary computer program and an input, whether the program will finish running, or continue to run forever. Alan Turing proved in 1936 that a general algorithm to solve the halting problem for all possible program–input pairs cannot exist. For any program f that might determine whether programs halt, a "pathological" program g, called with some input, can pass its own source and its input to f and then specifically do the opposite of what f predicts g will do. No f can exist that handles this case. A key part of the proof is a mathematical definition of a computer and program, which is known as a Turing machine; the halting problem is undecidable over Turing machines. It is one of the first cases of decision problems proven to be unsolvable. This proof is significant to practical computing efforts, defining a class of applications which no programming invention can possibly perform perfectly.
|
From the following paragraph, what is the main conclusion about the halting problem?
|
The halting problem is unsolvable.
|
null | false
| null |
Carl Rimmer (born 29 April 1986 in Banbury, England) is a Rugby Union player for Exeter Chiefs in the Aviva Premiership.
Rimmer is equally capable of playing at both loosehead and tighthead prop and made his debut for Exeter against London Welsh on 16 September 2012, subsequently starting at both loosehead and tighthead. Rimmer joined Exeter from fellow westcountry side Cornish Pirates and is qualified to play for both England or Wales. He was a replacement as Exeter Chiefs defeated Wasps to be crowned champions of the 2016-17 English Premiership.
|
Who is Carl Rimmer
|
Carl Rimmer (born 29 April 1986 in Banbury, England) is a Rugby Union player for Exeter Chiefs in the Aviva Premiership.
|
null | false
| 113
|
In this section we discuss the state of the art on conversational systems in three perspectives: types of interactions, types of architecture, and types of context reasoning. Then we present a table that consolidates and compares all of them.
ELIZA BIBREF11 was one of the first softwares created to understand natural language processing. Joseph Weizenbaum created it at the MIT in 1966 and it is well known for acting like a psychotherapist and it had only to reflect back onto patient's statements. ELIZA was created to tackle five "fundamental technical problems": the identification of critical words, the discovery of a minimal context, the choice of appropriate transformations, the generation of appropriate responses to the transformation or in the absence of critical words, and the provision of an ending capacity for ELIZA scripts.
Right after ELIZA came PARRY, developed by Kenneth Colby, who is psychiatrist at Stanford University in the early 1970s. The program was written using the MLISP language (meta-lisp) on the WAITS operating system running on a DEC PDP-10 and the code is non-portable. Parts of it were written in PDP-10 assembly code and others in MLISP. There may be other parts that require other language translators. PARRY was the first system to pass the Turing test - the psychiatrists were able to make the correct identification only 48 percent of the time, which is the same as a random guessing.
A.L.I.C.E. (Artificial Linguistic Internet Computer Entity) BIBREF12 appeared in 1995 but current version utilizes AIML, an XML language designed for creating stimulus-response chat robots BIBREF13 . A.L.I.C.E. bot has, at present, more than 40,000 categories of knowledge, whereas the original ELIZA had only about 200. The program is unable to pass the Turing test, as even the casual user will often expose its mechanistic aspects in short conversations.
Cleverbot (1997-2014) is a chatbot developed by the British AI scientist Rollo Carpenter. It passed the 2011 Turing Test at the Technique Techno-Management Festival held by the Indian Institute of Technology Guwahati. Volunteers participate in four-minute typed conversations with either Cleverbot or humans, with Cleverbot voted 59.3 per cent human, while the humans themselves were rated just 63.3 per cent human BIBREF14 .
Therefore, from the interaction point of view, there are two types: 1) one in which the chatbot was designed to chat with one person or chatbot, and 2) other in which the chatbot can interact with more than two members in the chat.
|
What are the two types of conversation from the interaction point of view?
|
There are two types: 1) one in which the chatbot was designed to chat with one person or chatbot, and 2) other in which the chatbot can interact with more than two members in the chat.
|
null | false
| null |
The history of ITV, the United Kingdom "Independent Television" commercial network, goes back to 1955.
Independent Television began as a network of independently-owned regional companies that were both broadcasters and programme makers, beginning with four companies operating six stations in three large regions in 1955–1956, and gradually expanding to 17 stations in 14 regions by 1962. Each regional station was responsible for its own branding, scheduling and advertising, with many peak-time programmes shared simultaneously across the whole network.
By 29 February 2016, 12 regions in England and Wales shared national ITV branding and scheduling, and, together with a 13th region UTV in Northern Ireland, were owned by a single company, ITV plc. A further two regions in Scotland carry STV branding and are owned by the STV Group.
|
When did ITV start broadcasting?
|
ITV first broadcast in 1955.
|
null | false
| null |
The B61 nuclear bomb is the primary thermonuclear gravity bomb in the United States Enduring Stockpile following the end of the Cold War. It is a low to intermediate-yield strategic and tactical nuclear weapon featuring a two-stage radiation implosion design.
|
What is the B61 nuclear bomb?
|
The B61 nuclear bomb is a low to intermediate yield strategic and tactical nuclear weapon. It features a two-stage radiation implosion design, and is the primary thermonuclear gravity bomb in the United States Enduring stockpile following the end of the Cold War.
|
null | false
| null |
How do people get where they want to go
|
People will typically look at a map and search for their destination. Then they follow the instructions on the map to get to their destination
|
|
null | false
| null |
Identify which instrument is string or percussion: Candombe drums, Datong
|
Datong is string, Candombe drums is percussion.
|
|
null | false
| 464
|
Our method is sensitive to the choice of basis elements. Selecting a maximum frequency that is too high leads to too many elements in the encoding and overfitting, while low frequencies tend to underfit. Similar limitations were observed by , where the "scale" parameter, which is related to the frequency of the basis elements, needs to be carefully tuned.
Evaluating the special functions that compose the basis used for positional encoding is generally computationally expensive. This is amortized when the inputs lie on a fixed grid (as in the spherical panoramas) or can be pre-computed from the dataset (like in the NeRF and light field experiments). This, combined with the fast on-device jax.scipy implementation of the spherical harmonics results in almost no overhead for these experiments. For IPDF, however, the training grid is sampled randomly, and there is no on-device implementation, which results is half the training speed. During inference the grid is fixed but very large, requiring a 32 Gb device for fast evaluation.
Our method is sensitive to the choice of basis elements. Selecting a maximum frequency that is too high leads to too many elements in the encoding and overfitting, while low frequencies tend to underfit. Similar limitations were observed by Tancik et al. (2020), where the “scale” parameter, which is related to the frequency of the basis elements, needs to be carefully tuned. Evaluating the special functions that compose the basis used for positional encoding is generally computationally expensive. This is amortized when the inputs lie on a fixed grid (as in the spherical panoramas) or can be pre-computed from the dataset (like in the NeRF (Mildenhall et al., 2020) and light field experiments). This, combined with the fast on-device jax.scipy implementation of the spherical harmonics results in almost no overhead for these experiments. For IPDF (Murphy et al., 2021), however, the training grid is sampled randomly, and there is no on-device implementation, which results is half the training speed. During inference the grid is fixed but very large, requiring a 32 Gb device for fast evaluation.
|
Can the author(s) quantify the increase in computational cost?
|
We discuss the computational cost in section 6. In summary, for the spherical panoramas, NeRF and light field experiments, all the points queried are known a priori and need to be computed only once. This, combined with the fast on-device jax.scipy spherical harmonics implementation, allows running these experiments with negligible overhead. However, we have no such tool for computing the Wigner-Ds for IPDF, and random points on SO(3) are queried during training so pre-computation is not possible. Our solution is to compute the basis elements on CPU and transfer to the device every step, which makes training 2x slower. This could be greatly improved with an on-device JAX Wigner-D implementation.
|
1910.11471
| false
| null |
SMT techniques require a parallel corpus in thr source and thr target language. A text-code parallel corpus similar to Fig. FIGREF12 is used in training. This parallel corpus has 18805 aligned data in it . In source data, the expression of each line code is written in the English language. In target data, the code is written in Python programming language.
In target data, the code is written in Python programming language.
|
What programming language is target language?
|
The answers are shown as follows:
* Python
|
null | false
| 125
|
Despite the American public's increasing acceptance of LGBTQ people and recent legal successes, LGBTQ individuals frequently remain the targets of hate and violence BIBREF0, BIBREF1, BIBREF2. At the core of this issue is dehumanization, “the act of perceiving or treating people as less than human” BIBREF3, a process that heavily contributes to extreme intergroup bias BIBREF4. Language is central to studying this phenomenon; like other forms of bias BIBREF5, BIBREF6, BIBREF7, BIBREF8, BIBREF9, dehumanizing attitudes are expressed through subtle linguistic manipulations, even in carefully-edited texts. It is crucial to understand the use of such linguistic signals in mainstream media, as the media's representation of marginalized social groups has far-reaching implications for social acceptance, policy, and safety.
While small-scale studies of dehumanization and media representation of marginalized communities provide valuable insights (e.g. BIBREF10), there exist no known large-scale analyses, likely due to difficulties in quantifying such a subjective and multidimensional psychological process. However, the ability to do large-scale analysis is crucial for understanding how dehumanizing attitudes have evolved over long periods of time. Furthermore, by being able to account for a greater amount of media discourse at once, large-scale techniques can provide a more complete view of the media environment to which the public is exposed.
Linguistics and computer science offer valuable methods and insights on which such large-scale techniques might be developed for the study of dehumanization. By leveraging more information about the contexts in which marginalized groups are discussed, computational linguistic methods not only enable large-scale study of a complex psychological phenomenon, but can even reveal linguistic variations and changes that are not easily identifiable through qualitative analysis alone.
In this work, we develop a computational linguistic framework for analyzing dehumanizing language, with a focus on lexical signals of dehumanization. Social psychologists have identified numerous components of dehumanization, such as negative evaluations of a target group, denial of agency, moral disgust, and likening members of a target group to non-human entities such as vermin. Drawing upon this rich body of literature, we first identify clear linguistic analogs for these components and propose several computational techniques to measure these linguistic correlates. We then apply this general framework to explore changing representations of LGBTQ groups in the New York Times over the span of three decades. We additionally use this lens of dehumanization to investigate differences in social meaning between the denotationally-similar labels gay and homosexual.
This paper aims to bridge the gaps between computational modeling, sociolinguistics, and dehumanization research with implications for several disciplines. In addition to enabling large-scale studies of dehumanizing language and media representation of marginalized social groups, these techniques can be built into systems that seek to capture both conscious and unconscious biases in text. Furthermore, this work has implications for improving machines' abilities to automatically detect hate speech and abusive language online, which are typically underpinned by dehumanizing language. Finally, our case study demonstrates that such computational analyses of discussions about marginalized groups can provide unique insights into language variation and change within sensitive sociopolitical contexts, and help us understand how people (and institutions) use language to express their ideologies and attitudes towards certain social groups.
Finally, our case study demonstrates that such computational analyses of discussions about marginalized groups can provide unique insights into language variation and change within sensitive sociopolitical contexts, and help us understand how people (and institutions) use language to express their ideologies and attitudes towards certain social groups.
|
What does their case study demonstrate?
|
Such computational analyses of discussions about marginalized groups can provide unique insights into language variation and change within sensitive sociopolitical contexts, and help them understand how people (and institutions) use language to express their ideologies and attitudes towards certain social groups.
|
null | false
| null |
India is a megadiverse country, a term employed for 17 countries which display high biological diversity and contain many species exclusively indigenous, or endemic, to them. India is a habitat for 8.6% of all mammal species, 13.7% of bird species, 7.9% of reptile species, 6% of amphibian species, 12.2% of fish species, and 6.0% of all flowering plant species. Fully a third of Indian plant species are endemic. India also contains four of the world's 34 biodiversity hotspots, or regions that display significant habitat loss in the presence of high endemism.
|
From the text below extract the types of species and the percentage of its habitat in the form {Species}:{Percentage} and place them as a bullet list
|
Here are the types of species and their percentages stated in the given text:
Mammal species : 8.6%
Bird species : 13.7%
Reptile species : 7.9%
Amphibian species : 6%
Fish species : 12.2%
Flowering Plant species : 6.0%
|
null | false
| null |
Several methods exist for storing hydrogen. These include mechanical approaches such as using high pressures and low temperatures, or employing chemical compounds that release H2 upon demand. While large amounts of hydrogen are produced by various industries, it is mostly consumed at the site of production, notably for the synthesis of ammonia. For many years hydrogen has been stored as compressed gas or cryogenic liquid, and transported as such in cylinders, tubes, and cryogenic tanks for use in industry or as propellant in space programs. Interest in using hydrogen for on-board storage of energy in zero-emissions vehicles is motivating the development of new methods of storage, more adapted to this new application. The overarching challenge is the very low boiling point of H2: it boils around 20.268 K (−252.882 °C or −423.188 °F). Achieving such low temperatures requires expending significant energy.
https://en.wikipedia.org/wiki/Hydrogen_storage
|
Given these paragraphs about Hydrogen Storage, what are some of its use cases?
|
It's been used in the industry or as propellant in space programs. Interest in using hydrogen for on-board storage of energy in zero-emissions vehicles is motivating the development of new methods of storage, more adapted to this new application.
|
1612.05310
| false
| null |
In order to provide directions for future work, we analyze the errors made by the classifier trained on the extended features on the four prediction tasks.
Errors on Intention (I) prediction: The lack of background is a major problem when identifying trolling comments. For example, “your comments fit well in Stormfront” seems inoffensive on the surface. However, people who know that Stormfront is a white supremacist website will realize that the author of this comment had an annoying or malicious intention. But our system had no knowledge about it and simply predicted it as non-trolling. These kind of errors reduces recall on the prediction of trolling comments. A solution would be to include additional knowledge from anthologies along with a sentiment or polarity. One could modify NELL BIBREF12 to broaden the understanding of entities in the comments.
Non-cursing aggressions and insults This is a challenging problem, since the majority of abusive and insulting comments rely on profanity and swearing. The problem arises with subtler aggressions and insults that are equally or even more annoying, such as “Troll? How cute.” and “settle down drama queen”. The classifier has a more difficult task of determining that these are indeed aggressions or insults. This error also decreases the recall of trolling intention. A solution would be to exploit all the comments made by the suspected troll in the entire conversation in order to increase the chances of finding curse words or other cues that lead the classifier to correctly classify the comment as trolling.
Another source of error is the presence of controversial topic words such as “black”,“feminism”, “killing”, “racism”, “brown”, etc. that are commonly used by trolls. The classifier seems too confident to classify a comment as trolling in the presence of these words, but in many cases they do not. In order to ameliorate this problem, one could create ad-hoc word embeddings by training glove or other type of distributed representation on a large corpus for the specific social media platform in consideration. From these vectors one could expect a better representation of controversial topics and their interactions with other words so they might help to reduce these errors.
Errors on Disclosure (D) prediction: A major source of error that affects disclosure is the shallow meaning representation obtained from the BOW model even when augmented with the distributional features given by the glove vectors. For example, the suspected troll's comment “how to deal with refugees? How about a bullet to the head” is clearly mean-spirited and is an example of disclosed trolling. However, to reach that conclusion the reader need to infer the meaning of “bullet to the head” and that this action is desirable for a vulnerable group like migrants or refugees. This problem produces low recall for the disclosed prediction task. A solution for this problem may be the use of deeper semantics, where we represent the comments and sentences in their logical form and infer from them the intended meaning.
Errors on Interpretation (R) prediction: it is a common practice from many users to directly ask the suspected troll if he/she is trolling or not. There are several variations of this question, such as “Are you a troll?” and “not sure if trolling or not”. While the presence of a question like these seems to give us a hint of the responder's interpretation, we cannot be sure of his interpretation without also considering the context. One way to improve interpretation is to exploit the response strategy, but the response strategy in our model is predicted independently of interpretation. So one solution could be similar to the one proposed above for the disclosure task problem: jointly learning classifiers that predict both variables simultaneously. Another possibility is to use the temporal sequence of response comments and make use of older response interpretation as input features for later comments. This could be useful since commenters seem to influence each other as they read through the conversation.
Errors on Response Strategy (B) prediction: In some cases there is a blurry line between “Frustrate” and “Neutralize”. The key distinction between them is that there exists some criticism in the Frustrate responses towards the suspected troll's comment, while “Neutralizing” comments acknowledge that the suspected troll has trolling intentions, but gives no importance to them. For example, response comments such as “oh, you are a troll” and “you are just a lame troll” are examples of this subtle difference. The first is a case of “neutralize” while the second is indeed criticizing the suspected troll's comment and therefore a “frustrate” response strategy. This kind of error affects both precision and recall for these two classes. A possible solution could be to train a specialized classifier to disambiguate between “frustrate” and “neutralize” only.
Another challenging problem is the distinction between the classes “Troll” and “Engage”. This is true when the direct responder is intensely flared up with the suspected comment to the point that his own comment becomes a trolling attempt. A useful indicator for distinguishing these cases are the presence of insults, and to detect them we look for swear words, but as we noted before, there is no guarantee that swear words are used for insulting. This kind of error affects the precision and recall for the “troll” and “engage” classes. A solution to this problem may be the inclusion of longer parts of the conversation. It is typical in a troll-engaged comment scheme to observe longer than usual exchanges between two users, and the comments evolve in very agitated remarks. One may then use this information to disambiguate between the two classes.
In order to provide directions for future work, we analyze the errors made by the classifier trained on the extended features on the four prediction tasks.
Errors on Intention (I) prediction: The lack of background is a major problem when identifying trolling comments.
Non-cursing aggressions and insults This is a challenging problem, since the majority of abusive and insulting comments rely on profanity and swearing.
Another source of error is the presence of controversial topic words such as “black”,“feminism”, “killing”, “racism”, “brown”, etc. that are commonly used by trolls.
Errors on Disclosure (D) prediction: A major source of error that affects disclosure is the shallow meaning representation obtained from the BOW model even when augmented with the distributional features given by the glove vectors.
Errors on Interpretation (R) prediction: it is a common practice from many users to directly ask the suspected troll if he/she is trolling or not.
Errors on Response Strategy (B) prediction: In some cases there is a blurry line between “Frustrate” and “Neutralize”.
Another challenging problem is the distinction between the classes “Troll” and “Engage”.
|
What is an example of a difficult-to-classify case?
|
The answers are shown as follows:
* The lack of background
* Non-cursing aggressions and insults
* the presence of controversial topic words
* shallow meaning representation
* directly ask the suspected troll if he/she is trolling or not
* a blurry line between “Frustrate” and “Neutralize”
* distinction between the classes “Troll” and “Engage”
|
null | false
| null |
Singapore (/ˈsɪŋ(ɡ)əpɔːr/ (listen)), officially the Republic of Singapore, is an island country and city-state in maritime Southeast Asia. It lies about one degree of latitude (137 kilometres or 85 miles) north of the equator, off the southern tip of the Malay Peninsula, bordering the Strait of Malacca to the west, the Singapore Strait to the south, the South China Sea to the east, and the Straits of Johor to the north. The country's territory is composed of one main island, 63 satellite islands and islets, and one outlying islet; the combined area of these has increased by 25% since the country's independence as a result of extensive land reclamation projects. It has the third highest population density in the world. With a multicultural population and recognising the need to respect cultural identities of the major ethnic groups within the nation, Singapore has four official languages: English, Malay, Mandarin, and Tamil. English is the lingua franca and numerous public services are available only in English. Multi-racialism is enshrined in the constitution and continues to shape national policies in education, housing, and politics.
|
What is the official name of Singapore?
|
The official name of Singapore is The Republic of Singapore
|
null | false
| null |
Classify the below companies based on their market capitalization into Small Cap and Large Cap.
Gravita, MapmyIndia, Airtel, Carysil
|
Small Cap - Gravita, MapmyIndia, Carysil
Large Cap - Airtel
|
|
null | false
| null |
Choose the word which is different from the rest: Mother, Friend, Sister, Father, Brother
|
Friend is a term reflecting a bond that is not genetic
|
|
null | false
| 154
|
Four publicly available corpora are selected for our analysis. These corpora are based on Wikipedia, so more comparable than the others, and have already been used for the evaluation of several QA systems.
WikiQA BIBREF6 comprises questions selected from the Bing search queries, where user click data give the questions and their corresponding Wikipedia articles. The abstracts of these articles are then extracted to create answer candidates. The assumption is made that if many queries lead to the same article, it must contain the answer context; however, this assumption fails for some occasions, which makes this dataset more challenging. Since the existence of answer contexts is not guaranteed in this task, it is called answer triggering instead of answer selection.
SelQA BIBREF7 is a product of five annotation tasks through crowdsourcing. It consists of about 8K questions where a half of the questions are paraphrased from the other half, aiming to reduce contextual similarities between questions and answers. Each question is associated with a section in Wikipedia where the answer context is guaranteed, and also with five sections selected from the entire Wikipedia where the selection is made by the Lucene search engine. This second dataset does not assume the existence of the answer context, so can be used for the evaluation of answer triggering.
SQuAD BIBREF12 presents 107K+ crowdsourced questions on 536 Wikipedia articles, where the answer contexts are guaranteed to exist within the provided paragraph. It contains annotation of answer phrases as well as the pointers to the sentences including the answer phrases; thus, it can be used for both answer extraction and selection. This corpus also provides human accuracy on those questions, setting up a reasonable upper bound for machines. To avoid overfitting, the evaluation set is not publicly available although system outputs can be evaluated by their provided script.
InfoboxQA BIBREF13 gives 15K+ questions based on the infoboxes from 150 articles in Wikipedia. Each question is crowdsourced and associated with an infobox, where each line of the infobox is considered an answer candidate. This corpus emphasizes the gravity of infoboxes, which summary arguably the most commonly asked information about those articles. Although the nature of this corpus is different from the others, it can also be used to evaluate answer selection.
It contains annotation of answer phrases as well as the pointers to the sentences including the answer phrases; thus, it can be used for both answer extraction and selection.
|
Why the SQUAD can be used for both answer extraction and selection?
|
Because it contains annotation of answer phrases as well as the pointers to the sentences including the answer phrases.
|
null | false
| 391
|
Most of our parameter settings follow BIBREF13 and BIBREF6 . We use ( 29 ) to update the model, and use ADAM BIBREF25 with a learning rate of 0.001 and batch size of 64 for optimization. A small dropout rate of 0.1 is applied to each layer. For word embedding, we use 300-dimension GloVe BIBREF21 embedding from the 840B Common Crawl corpus. The word embeddings are not updated during training. The character embedding has 20 dimensions and the character LSTM has 50 hidden units. All other LSTMs have a hidden dimension of 100. The maximum reasoning step $\mathcal {T}$ is set to 5. We limit the length of passage/question/answer to a maximum of 500/100/100 for efficient computation. We also train an ensemble model of 9 DFNs using randomly initialized parameters. Training usually converges within 20 epochs. The model is implemented with Tensorflow BIBREF26 and the source code will be released upon paper acceptance.
Most of our parameter settings follow (Wang et al., 2017c) and (Shen et al., 2017). We use (1) to update the model, and use ADAM (Kingma and Ba, 2014) with a learning rate of 0.001 and batch size of 64 for optimization.
|
How much is their learning rate?
|
0.001.
|
null | false
| null |
Is toastmasters a good organization to join?
|
Toastmasters is an organization that has been around for nearly 100 years. It is an organization that helps its members to become competent communicators and competent leaders. They achieve this by delivering prepared speeches that are evaluated during Toastmasters meetings. You get evaluated on filler words used, grammatical construction of your speech, speech tonality, and a general evaluation of your speech. This works so well because you get practical insight into where you can improve in your communication skills in delivering prepared speeches. You can also fill different roles like treasurer, and president roles to help you build leadership skills.
|
|
2003.05995
| false
| null |
Wizard interface: the interface shown to participants with the Wizard role provides possible actions on the right-hand side of the browser window. These actions could be verbal, such as sending a message, or non-verbal, such as switching on/off a button to activate a robot. Figure FIGREF11 shows this interface with several actions available to be used in our data collection.
Dialogue structure: we introduced structured dialogues through a Finite State Machine (FSM) that controls the current dialogue state and offers multiple suitable and relevant state transitions (actions) to the Wizard depending on the point in the interaction, the state of the world and the history. A graph of dialogue states, transitions and utterances is loaded when the system is initialised, and each chat room has its own dialogue state, which changes through actions.
System-changing actions: actions trigger transitions between the states in the FSM. We differentiate two types of actions:
Verbal actions, such as the dialogue options available at that moment. The Wizard can select one of several predefined messages to send, or type their own message if needed. Free text messages do not change the dialogue state in the FSM, so it is important to minimise their use by providing enough dialogue options to the Wizard. Predefined messages can also trigger other associated events such as pop-ups or follow-up non-verbal actions.
Non-verbal actions, such as commands to trigger events. These can take any form, but we used buttons to control robots in our data collection.
Submitting an action would change the dialogue state in the FSM, altering the set of actions available in the subsequent turn visible to the Wizard. Some dialogue options are only possible at certain states, in a similar way as to how non-verbal actions are enabled or disabled depending on the state. This is reflected in the Wizard interface.
Wizard interface: the interface shown to participants with the Wizard role provides possible actions on the right-hand side of the browser window. These actions could be verbal, such as sending a message, or non-verbal, such as switching on/off a button to activate a robot.
Dialogue structure: we introduced structured dialogues through a Finite State Machine (FSM) that controls the current dialogue state and offers multiple suitable and relevant state transitions (actions) to the Wizard depending on the point in the interaction, the state of the world and the history. A graph of dialogue states, transitions and utterances is loaded when the system is initialised, and each chat room has its own dialogue state, which changes through actions.
System-changing actions: actions trigger transitions between the states in the FSM. We differentiate two types of actions:
Verbal actions, such as the dialogue options available at that moment. The Wizard can select one of several predefined messages to send, or type their own message if needed. Free text messages do not change the dialogue state in the FSM, so it is important to minimise their use by providing enough dialogue options to the Wizard. Predefined messages can also trigger other associated events such as pop-ups or follow-up non-verbal actions.
Non-verbal actions, such as commands to trigger events. These can take any form, but we used buttons to control robots in our data collection.
Submitting an action would change the dialogue state in the FSM, altering the set of actions available in the subsequent turn visible to the Wizard. Some dialogue options are only possible at certain states, in a similar way as to how non-verbal actions are enabled or disabled depending on the state.
|
How does framework made sure that dialogue will not breach procedures?
|
The answers are shown as follows:
* The Wizard can select one of several predefined messages to send, or type their own message if needed. Free text messages do not change the dialogue state in the FSM, so it is important to minimise their use by providing enough dialogue options to the Wizard. Predefined messages can also trigger other associated events such as pop-ups or follow-up non-verbal actions.
|
null | false
| null |
Choose the word which is different from the rest: Barley, Rice, Wheat, Mustard, Bajra.
|
All except Mustard are foodgrains, while mustard is an oilseed. A grain is a small, hard, dry fruit with or without an attached hull layer, harvested for human or animal consumption.
The major U.S. oilseed crops are soybeans, cottonseed, sunflower seed and canola. Soybeans are the dominant oilseed in the United States, accounting for about 90 percent of U.S. oilseed production.
|
|
1910.11235
| false
| null |
In this paper, we adopt two simple strategies, multi-range reinforcing and multi-entropy sampling to overcome the reward sparseness during training. With the tricks applied, our model demonstrates a significant improvement over competing models. In addition, we propose road exam as a new metric to reveal a model's robustness against exposure bias.
In addition, we propose road exam as a new metric to reveal a model's robustness against exposure bias.
|
What is the road exam metric?
|
The answers are shown as follows:
* a new metric to reveal a model's robustness against exposure bias
|
null | false
| null |
Who is the white rapper that sings 'Thrift Shop'?
|
macklemore
|
|
null | false
| null |
The FIVB Volleyball Men's Club World Championship is an international men's club volleyball competition organised by the Fédération Internationale de Volleyball (FIVB), the sport's global governing body. The competition was first contested in 1989 in Italy. It was not held between 1993 and 2008, but since 2009, the competition has been held every year, and has been hosted by Qatar and Brazil except in 2017 and 2018 where both hosted by Poland.
|
Since it began, how many years has FIVB not been held?
|
16 years. 2008-1993+1 = 16.
|
null | false
| null |
Where are the headquarters of FATF
|
FATF has headquarters in Paris, France
|
|
null | false
| null |
What is the SECA Art Award?
|
SECA stands for Society for the Encouragement of Contemporary Art. The SECA Art Award is an annual award that is given to emerging San Francisco Bay Area artists. It was organized by the SFMOMA in1967, and has been awarded every year since then.
In 2022, 5 artists got selected: Binta Ayofemi, Maria A. Guzmán Capron, Cathy Lu, Marcel Pardo Ariza, and Gregory Rick. Their diverse works deal with racial conflict, American Asian identity, trans pride, black and indigenous presence in the urban landscape.
SECA installations can be viewed at the SFMOMA Level 2 galleries from December through May annually.
|
|
null | false
| null |
What is the best way to drink coffee?
|
The best way to drink coffee is subjective. Everyone likes their coffee a different way. Some prefer iced coffee. Some people prefer hot coffee. Some people like their coffee with cream and sugar. Some people like their coffee black. Some people get their coffee from Starbucks and some people make their coffee at home.
|
|
1611.03382
| false
| null |
Evaluation on DUC2004: DUC 2004 ( BIBREF15 ) is a commonly used benchmark on summarization task consisting of 500 news articles. Each article is paired with 4 different human-generated reference summaries, capped at 75 characters. This dataset is evaluation-only. Similar to BIBREF2 , we train our neural model on the Gigaword training set, and show the models' performances on DUC2004. Following the convention, we also use ROUGE limited-length recall as our evaluation metric, and set the capping length to 75 characters. We generate summaries with 15 words using beam-size of 10. As shown in Table TABREF35 , our method outperforms all previous methods on Rouge-1 and Rouge-L, and is comparable on Rouge-2. Furthermore, our model only uses 15k decoder vocabulary, while previous methods use 69k or 200k.
FLOAT SELECTED: Table 2: Rouge-N limited-length recall on DUC2004. Size denotes the size of decoder vocabulary in a model.
As shown in Table TABREF35 , our method outperforms all previous methods on Rouge-1 and Rouge-L, and is comparable on Rouge-2.
FLOAT SELECTED: Table 2: Rouge-N limited-length recall on DUC2004. Size denotes the size of decoder vocabulary in a model.
|
By how much does their model outperform both the state-of-the-art systems?
|
w.r.t Rouge-1 their model outperforms by 0.98% and w.r.t Rouge-L their model outperforms by 0.45%
|
null | false
| null |
Who are the three main characters of the Harry Potter Series?
|
Harry Potter, Hermione Granger and Ron Weasley are the main characters of Harry Potter
|
|
null | false
| null |
Microsoft Corporation is an American multinational technology corporation headquartered in Redmond, Washington. Microsoft's best-known software products are the Windows line of operating systems, the Microsoft Office suite, and the Internet Explorer and Edge web browsers. Its flagship hardware products are the Xbox video game consoles and the Microsoft Surface lineup of touchscreen personal computers. Microsoft ranked No. 14 in the 2022 Fortune 500 rankings of the largest United States corporations by total revenue; it was the world's largest software maker by revenue as of 2022. It is considered as one of the Big Five American information technology companies, alongside Alphabet (parent company of Google), Amazon, Apple, and Meta (formerly Facebook).
|
From this paragraph about Microsoft, extract the names of the Big Five American information technology companies.
|
1) Microsoft, 2) Alphabet (parent company of Google), Amazon, Apple, and Meta (formerly Facebook).
|
null | false
| null |
West's outspoken views have received significant media coverage; he has been a frequent source of controversy due to his conduct on social media and at awards shows and public settings, as well as his comments on the music and fashion industries, U.S. politics, race, and slavery. His Christian faith, high-profile marriage to Kim Kardashian, and mental health have also been topics of media attention. In 2020, West launched an unsuccessful independent presidential campaign that primarily advocated for a consistent life ethic. In 2022, he was widely condemned and lost many sponsors and partnerships—including his collaborations with Adidas, Gap, and Balenciaga—after making a series of antisemitic statements. In November 2022, he announced his 2024 presidential campaign, appearing publicly with Nick Fuentes, a white supremacist. West later publicly praised Adolf Hitler, denied the Holocaust, and identified as a Nazi.
|
Why has West been a source of controversy recently?
|
West began making antisemitic comments on Twitter and on other news sources in 2022. His comments directly attacked Jewish people and denounced parts of history such as the Holocaust. Once he made his comments, many organizations that he was partnered with dropped him because of the backlash received by their customers.
|
null | false
| null |
Identify the bird from the list: Barn owl, Python, Chimpanzee
|
Barn owl
|
|
null | false
| null |
Jimmy John's is an American sandwich chain headquartered in Champaign, Illinois. The business was founded by Jimmy John Liautaud in 1983. After Liautaud graduated from high school, his father gave him a choice to either join the military or start a business. Liautaud, choosing the latter, initially decided to start a hot dog business, but changed to opening a sandwich business due to costs.
The Jimmy John's franchise has over 2700 locations, with 98% of them being franchises. in 2016, Roark Capital Group purchased a majority stake in the company. Later, in 2019, Inspire Brands purchased the company from Roark Capital Group for an unspecified amount.
|
Given this reference text about Jimmy John's, how many Jimmy John's locations are there?
|
There are currently 2700 locations in the Jimmy John's franchise.
|
null | false
| null |
Nokia is a public limited liability company and is the oldest company listed under the same name on the Helsinki Stock Exchange, beginning in 1915. Nokia has had a secondary listing on the New York Stock Exchange since 1994. Nokia shares were delisted from the London Stock Exchange in 2003, the Paris Stock Exchange in 2004, the Stockholm Stock Exchange in 2007 and the Frankfurt Stock Exchange in 2012. Due to the acquisition of Alcatel-Lucent in 2015, Nokia listed its shares again on the Paris Stock Exchange and was included in the CAC 40 index on 6 January 2016 but later removed on 18 September 2017.
In 2007, Nokia had a market capitalization of €110 billion; by 17 July 2012 this had fallen to €6.28 billion, and by 23 February 2015, it increased to €26.07 billion. Nokia market cap at 2020 was 21.76 billion.
|
From the text, please extract and list Nokia's market capitalization timeline.
|
Here is a timeline of Nokia's market capitalization:
1. 2007 - 110 Billion Euros
2. 2012 - 6.28 Billion Euros
3. 2015 - 26.07 Billion Euros
4. 2020 - 21.76 Billion Euros
|
null | false
| 125
|
Our lexical semantic analysis involves quantifying linguistic correlates of component psychological processes that contribute to dehumanization. Our approaches are informed by social psychology research on dehumanization, which is briefly summarized here. Prior work has identified numerous related processes that comprise dehumanization BIBREF4. One such component is likening members of the target group to non-human entities, such as machines or animals BIBREF4, BIBREF11, BIBREF12. By perceiving members of a target group to be non-human, they are “outside the boundary in which moral values, rules, and considerations of fairness apply" BIBREF13, which thus leads to violence and other forms of abuse. Metaphors and imagery relating target groups to vermin are particularly insidious and played a prominent role in the genocide of Jews in Nazi Germany and Tutsis in Rwanda BIBREF14. More recently, the vermin metaphor has been invoked by the media to discuss terrorists and political leaders of majority-Muslim countries after September 11 BIBREF15. According to BIBREF16, the vermin metaphor is particularly powerful because it conceptualizes the target group as “engaged in threatening behavior, but devoid of thought or emotional desire".
Disgust underlies the dehumanizing nature of these metaphors and is itself another important element of dehumanization. Disgust contributes to members of target groups being perceived as less-than-human and of negative social value BIBREF17. It is often evoked (both in real life and experimental settings) through likening a target group to animals. BIBREF18 find that priming participants to feel disgust facilitates “moral exclusion of out-groups". Experiments by BIBREF17 and BIBREF19 similarly find that disgust is a predictor of dehumanizing perceptions of a target group. Both moral disgust towards a particular social group and the invocation of non-human metaphors are facilitated by essentialist beliefs about groups, which BIBREF4 presents as a necessary component of dehumanization. In order to distinguish between human and non-human, dehumanization requires an exaggerated perception of intergroup differences. Essentialist thinking thus contributes to dehumanization by leading to the perception of social groups as categorically distinct, which in turn emphasizes intergroup differences BIBREF4
According to BIBREF4, much of the prior research describes “extremely negative evaluations of others" as a major component of dehumanization. This is especially pronounced in BIBREF20's account of delegitimization, which involves using negative characteristics to categorize groups that are “excluded from the realm of acceptable norms and values" BIBREF20. While BIBREF20 defines delegitimization as a distinct process, he considers dehumanization to be one means of delegitimization. BIBREF13 also discusses broader processes of moral exclusion, one of which is dehumanization. A closely related process is psychological distancing, in which one perceives others to be objects or nonexistent BIBREF13. BIBREF21 identifies elements that contribute to the objectification (and thus dehumanization) of women, one of which is denial of subjectivity, or the habitual neglect of one's experiences, emotions, and feelings.
Another component of dehumanization is the denial of agency to members of the target group BIBREF4. According to BIBREF16, there are three types of agency: "the ability to (1) experience emotion and feel pain (affective mental states), (2) act and produce an effect on their environment (behavioral potential), and (3) think and hold beliefs (cognitive mental states) BIBREF16. Dehumanization typically involves the denial of one or more of these types of agency BIBREF16.
In Section SECREF4, we introduce computational linguistic methods to quantify several of these components of dehumanization.
In Section SECREF1, we briefly discussed multiple elements of dehumanization that have been identified in social psychology literature. Here we introduce and quantify lexical correlates to operationalize four of these components: negative evaluations of a target group, denial of agency, moral disgust, and use of non-human metaphors (particularly vermin).
One prominent aspect of dehumanization is extremely negative evaluations of members of a target group BIBREF4. Attribution of negative characteristics to members of a target group in order to exclude that group from “the realm of acceptable norms and values" is specifically the key component of delegitimization, a process of moral exclusion closely related to dehumanization. We hypothesize that this negative evaluation of a target group can be realized by words and phrases whose connotations have extremely low valence, where the valence of a word refers to the dimension of meaning corresponding to positive/negative (or pleasure/displeasure) BIBREF55, BIBREF56. Thus, we propose several valence lexicon-based approaches to measure this component: paragraph-level valence analysis, Connotation Frames, and word embedding neighbor valence. Each of these techniques has different advantages and disadvantages regarding precision and interpretability.
As we mentioned above, one important dimension of affective meaning is valence, which corresponds to an individual's evaluation of an event or concept, ranging from negative/unpleasant to positive/pleasant BIBREF55, BIBREF57. A straightforward lexical approach to quantify negative evaluations of a target group is thus to measure the average valence over all words in corpora containing discussions of the target group. We quantify this by using the valence dimension from the NRC VAD lexicon, which contains real-valued scores ranging from zero to one for valence, arousal and dominance for 20,000 English words, where a score of zero is the lowest valence (most negative emotion) and a score of one is the highest possible valence (most positive emotion) BIBREF56. Words with the highest valence in the NRC VAD Lexicon include love, happy, and happily, while words with the lowest valence include toxic, nightmare, and shit.
In our case study, we qualitatively found that paragraphs were the optimal level of context for analysis; full articles often focused on topics unrelated to LGBTQ communities, even if an LGBTQ label was mentioned somewhere in the article, while single sentences did not provide enough discussion to get a sense of how LGBTQ groups are represented by the newspaper. Thus, we calculate paragraph-level scores by taking the average valence score over all words in the paragraph that appear (or whose lemmas appear) in the NRC VAD lexicon.
While paragraph-level valence analysis is quick and simple, it is sometimes too coarse because we aim to understand the sentiment directed towards the target group, not just nearby in the text. For example, suppose the target group is named “B". A sentence such as “A violently attacked B" would likely have extremely negative valence, but the writer may not feel negatively towards the victim, “B".
We address this by using BIBREF22's Connotation Frames Lexicon, which contains rich annotations for 900 English verbs (BIBREF22). Among other things, for each verb, the Connotation Frames Lexicon provides scores (ranging from -0.87 to 0.8) for the writer's perspective towards the verb's subject and object. In the example above for the verb attack, the lexicon lists the writer's perspective towards the subject “A", the attacker, as -0.6 (strongly negative) and the object “B" as 0.23 (weakly positive).
We extract all subject-verb-object tuples containing at least one target group label using the Spacy dependency parser . For each subject and object, we capture the noun and the modifying adjectives, as group labels (such as gay) can often take either nominal or adjectival forms. For each tuple, we use the connotation frame lexicon to determine the writer's perspective either towards the subject if the group label appears in the subject noun phrase, or perspective towards the object if the label appears in the object noun phrase. We then average perspective scores over all tuples.
While a Connotation Frames approach can be more precise than word-counting valence analysis, it limits us to analyzing SVO triples, which excludes a large portion of the available data about the target groups. This reveals a conundrum: broader context can provide valuable insights into the implicit evaluations of a social group, but we also want to directly probe attitudes towards the group itself.
We address this tension by training vector space models to represent the data, in which each unique word in a large corpus is represented by a vector (embedding) in high-dimensional space. The geometry of the resulting vector space captures many semantic relations between words. Furthermore, prior work has shown that vector space models trained on corpora from different time periods can capture semantic change BIBREF58, BIBREF59. For example, diachronic word embeddings reveal that the word gay meant “cheerful" or “dapper" in the early 20th century, but shifted to its current meaning of sexual orientation by the 1970s. Because word embeddings are created from real-world data, they contain real-world biases. For example, BIBREF37 demonstrated that gender stereotypes are deeply ingrained in these systems. Though problematic for the widespread use of these models in computational systems, these revealed biases indicate that word embeddings can actually be used to identify stereotypes about social groups and understand how they change over time BIBREF24.
This technique can similarly be applied to understand how a social group is negatively evaluated within a large text corpus. If the vector corresponding to a social group label is located in the semantic embedding space near negative words or terms with clearly negative evaluations, that group is likely negatively evaluated (and possibly dehumanized) in the text. These vector space models can be used in a diachronic study to capture subtle changes in the connotations of a group label over time, even when the denotational meaning of the label has remained constant. Furthermore, comparing differences in what words are most closely associated with related group labels can reveal meaning variation between these labels, even if they refer to similar populations.
We first preprocess the data by lowercasing, removing numbers, and removing punctuation. We then use the word2vec skip-gram model to create word embeddings BIBREF60. The resulting vector spaces have 100 dimensions, and we specify the following additional parameters to the model: context window of size 10, minimum frequency of 5 for a word to be included, 10 negative samples, 10 iterations, and a sampling rate of $10^{-4}$. We then use cosine similarity to determine a group label's nearest neighbors in the vector space. For our diachronic analysis, we first train word2vec on the entire corpus, and then use the resulting vectors to initialize word2vec models for each year of data in order to encourage coherence and stability across years. After training word2vec, we zero-center and normalize all embeddings to alleviate the hubness problem BIBREF61.
We then identify vectors for group labels by taking the centroid of all morphological forms of the label, weighted by frequency. For example, the vector representation for the label gay is actually the weighted centroid of the words gay and gays. This enables us to simultaneously account for adjectival, singular nominal, and plural nominal forms for each social group label with a single vector. Finally, we estimate the valence for each group label by identifying its 1000 nearest neighbors via cosine similarity (1000 was chosen arbitrarily), and calculating the average valence of all neighbors that appear in the NRC VAD valence lexicon.
We also tried using a group label's vector representation to induce a valence score directly, instead of using nearest neighbors as a proxy, by adapting the regression-based sentiment prediction from BIBREF62 for word embeddings. We found that this approach yielded similar results as analyzing nearest neighbor valence but was difficult to qualitatively interpret. More details for and results from this technique can be found in the Appendix.
Denial of agency refers to the lack of attributing a target group member with the ability to control their own actions or decisions BIBREF16. Automatically detecting the extent to which a writer attributes cognitive abilities to a target group member is an extraordinarily challenging computational task. Fortunately, the same lexicons used to operationalize negative evaluations provide resources for measuring lexical signals of denial of agency.
As in Section SECREF6, we use Connotation Frames to quantify the amount of agency attributed to a target group. We use BIBREF25's extension of Connotation Frames for agency BIBREF25. Following BIBREF25's interpretation, entities with high agency exert a high degree of control over their own decisions and are active decision-makers, while entities with low agency are more passive BIBREF25. This contrast is particularly apparent in example sentences such as X searched for Y and X waited for Y, where the verb searched gives X high agency and waited gives X low agency BIBREF25. Additionally, BIBREF25's released lexicon for agency indicates that the subjects of verbs such as attack and praise are given high agency, while the subjects of doubts, needs, and owes are given low agency BIBREF25.
This lexicon considers the agency attributed to subjects of nearly 2000 transitive and intransitive verbs. To use this lexicon to quantify denial of agency in our corpus, we extract all sentences' head verbs and their corresponding subjects, where the subject noun phrase contains a target group label. Unlike BIBREF22's real-valued Connotation Frames lexicon for perspective, the agency lexicon only provides binary labels, so we calculate the fraction of subject-verb pairs where a subject containing a group label was given high agency by its head verb.
The NRC VAD lexicon's dominance annotations provide another resource for quantifying dehumanization BIBREF56. As with valence, the NRC VAD lexicon's dominance dimension contains real-valued scores between zero and one for 20,000 English words. However, it is important to note that the dominance lexicon primarily captures power, which is distinct from but closely related to agency. While power refers to one's control over others, agency refers to one's control over their own self. While this lexicon is a proxy, it qualitatively appears to capture signals of denial of agency; the highest dominance words are powerful, leadership, success, and govern, while the lowest dominance words are weak, frail, empty, and penniless. We thus take the same approach as in Section SECREF10 with the same vector space models. However, we now calculate the average dominance score of the 1000 nearest neighbors to each group label vector representation.
As in Section SECREF10, we also induced a dominance score directly from a group label's vector representation by adapting the regression-based sentiment prediction from BIBREF62 for word embeddings. More details and results for this technique can be found in the Appendix.
In order to operationalize the moral disgust component of dehumanization with lexical techniques, we draw inspiration from Moral Foundations theory, which postulates that there are five dimensions of moral intuitions: care, fairness/proportionality, loyalty/ingroup, authority/respect, and sanctity/purity BIBREF63. The negative end of the sanctity/purity dimension corresponds to moral disgust. While we do not directly incorporate Moral Foundations Theory in our framework for dehumanization, we utilize lexicons created by BIBREF64 corresponding to each moral foundation. The dictionary for moral disgust includes over thirty words and stems, including disgust*, sin, pervert, and obscen* (the asterisks indicate that the dictionary includes all words containing the preceding prefix, such as both obscene and obscenity).
We opt for a vector approach instead of counting raw frequencies of moral disgust-related words because words that explicitly invoke moral disgust are very sparse in our news corpus. Furthermore, vectors are capable of capturing associations with the social group label itself, while counts in paragraphs and full news articles would not directly capture such associations. Using the word embeddings from Section SECREF10, we construct a vector to represent the concept of moral disgust, which is the average of the corresponding vectors for all words in the “Moral Disgust" dictionary, weighted by word frequency. This method of creating a vector from the Moral Foundations dictionary resembles that used by BIBREF65. We can then analyze implicit associations between a social group and moral disgust by calculating the cosine similarity between the group label's vector and the Moral Disgust concept vector, where a higher similarity suggests closer associations between the social group and moral disgust. As previously, each group label's vector is the average of its morphological variants' vectors, weighted by frequency.
Metaphors comparing humans to vermin have been especially prominent in dehumanizing groups throughout history BIBREF4, BIBREF15. Even if a marginalized social group is not directly equated to vermin in the press, this metaphor may be invoked in more subtle ways, such as through the use of verbs that are also associated with vermin (like scurry as opposed to the more neutral hurry) BIBREF66. While there is some natural language processing work on the complex task of metaphor detection (e.g. BIBREF67, BIBREF67), such systems would not be able to easily quantify such indirect associations.
We thus quantify the metaphorical relationship between a social group and vermin by calculating similarities between these concepts in a distributional semantic vector space, as we did with moral disgust in Section SECREF14. We create a Vermin concept vector by calculating the average of the following vermin words' vectors, weighted by frequency: vermin, rodent(s), rat(s) mice, cockroaches, termite(s), bedbug(s), fleas . We do not include the singular mouse or flea because non-vermin senses of those words were more frequent (computer mouse and flea market, respectively), and word2vec does not account for polysemy. We calculate cosine similarity between each social group label and the Vermin concept vector. A large cosine similarity between a label and the Vermin vector suggests that the social group is closely associated with vermin.
Table TABREF18 provides an overview of the four elements of dehumanization that we study and the lexical techniques used to quantify them.
Our approaches are informed by social psychology research on dehumanization, which is briefly summarized here.
|
Their approaches are informed by what?
|
Social psychology research on dehumanization.
|
null | false
| null |
Identify which instrument is string or woodwind: Zither, Khlui
|
Zither is string, Khlui is woodwind.
|
|
null | false
| null |
Which is a species of fish? Sleeper or Snoozer
|
Sleeper
|
|
null | false
| null |
How many colors are there in a rainbow.
|
Seven
|
|
null | false
| null |
Is SAS a language or a software product?
|
SAS is both a programming language and a software product.
|
|
null | false
| null |
From Sayo Junction to Sayo Toll Gate the expressway is maintained and tolled by the West Nippon Expressway Company.
|
Who maintains the Tottori Expressway?
|
The expressway is officially referred to as the Chūgoku-Ōdan Expressway Himeji Tottori Route. The Chūgoku-Ōdan Expressway Himeji Tottori Route is the official designation for the Sanyō Expressway between Sanyō Himeji-Nishi Interchange and Harima Junction, the Harima Expressway between Harima Junction and Yamazaki Junction, the Chūgoku Expressway between Yamazaki Junction and Sayō Junction, and the Tottori Expressway between Sayō Junction and Tottori Interchange (concurrent with the Chūgoku-Ōdan Expressway Himeji Tottori Route).
From Sayo Junction to Sayo Toll Gate the expressway is maintained and tolled by the West Nippon Expressway Company. The rest of the expressway is able to be driven without any fees. That section of the expressway is maintained by the Chūgoku branch of the Ministry of Land, Infrastructure, Transport and Tourism.
The entire expressway has only one lane in each direction, except for the section between Sayō Junction and Sayō Toll Gate.
|
null | false
| null |
What's the difference between tennis and pickleball?
|
While they are similar sports, tennis and pickleball have various difference. First, the court size for pickleball is about half the size of a tennis court. Second, pickleball is played with a ball that resembles a whiffle ball. Third, pickleball is played with paddles as opposed to rackets. Finally, the scoring system is quite different as you play for points which can only be scored when you or your team are serving.
|
|
null | false
| null |
What is the trias politica?
|
The trias politica is the typical separation of powers in a government - the legislative power, the executive power and the judiciary power. These powers should be independent of each other, with each performing a different function, keeping each other in check. The legislative power drafts laws and regulations that govern the society. The executive power ensures these laws and regulations are upheld in society, while the judiciary power checks the performance of the executive power in upholding those laws and regulations.
|
|
null | false
| null |
Two small communities are separated by a distance of three miles. In the absence of a modern telecommunications infrastructure, think of five ways for inhabitants to efficiently exchange messages.
|
The first method that comes to mind involves physically carrying messages on foot or in a vehicle of some sort, and relaying them verbally or in written form.
An alternative approach still involving physical delivery of the message could be to employ some physical system to convey the messages for you. Examples of this principle include zip lines or pneumatic tubes.
A different approach could be to transmit the message acoustically, which at this distance would require a very loud signalling device, the assistance of a transmission medium of some sort, or repeaters along the way.
Encoding the information of the message and transmitting it visually provides yet a different approach. This could be in the form of light flashing or smoke signals for example.
Finally, employing some rudimentary physical system to transmit encoded information could be used. For example, a taut cable or hydraulic system whose changes or movements could be translated into information.
|
|
null | false
| null |
Classify each of the following as a planet or a star: Neptune, Saturn, Pollux, Sun, Jupiter, Mercury
|
Neptune: Planet
Saturn: Planet
Pollux: Star
Sun: Star
Jupiter: Planet
Mercury: Planet
|
|
null | false
| null |
How would you categorize the following sports into two groups; football, rugby, cricket and baseball
|
If I were to categorize these sports into 2 groups I would split them up into striking and invasion-based games. Football and Rugby are both invasion games, meaning two teams must attack their opposition's territory in order to score points. Baseball and cricket both involve two teams in set positions who must strike a ball in order to score points.
|
|
null | false
| null |
What were the last ten states to enter the union of the United States?
|
In order of admission: Montana, Washington, Idaho, Wyoming, Utah, Oklahoma, New Mexico, Arizona, Alaska, and Hawaii.
|
|
null | false
| 341
|
Recent advances in Machine Learning (ML) and, in particular, in Deep Neural Networks (DNN) have allowed the development of highly accurate predictive systems for numerous applications. Among others, health has received significant attention due to the potential of ML-based diagnostic, monitoring and therapeutic systems, which are fast (when compared to traditional diagnostic processes), easily distributed and cheap to implement (many such systems can be executed in mobile devices). Furthermore, these systems can incorporate biometric data to perform non-invasive diagnostics.
Among other data types, speech has been proposed as a valuable biomarker for the detection of a myriad of diseases, including: neurological conditions, such as Alzheimer’s BIBREF0, Parkinson’s disease (PD) BIBREF1 and Amyotrophic Lateral Sclerosis BIBREF2; mood disorders, such as depression, anxiety BIBREF3 and bipolar disorder BIBREF4; respiratory diseases, such as obstructive sleep apnea (OSA) BIBREF5. However, temporal and financial constraints, lack of awareness in the medical community, ethical issues and patient-privacy laws make the acquisition of medical data one of the greatest obstacles to the development of health-related speech-based classifiers, particularly for deep learning models. For this reason, most systems rely on knowledge-based (KB) features, carefully designed and selected to model disease symptoms, in combination with simple machine learning models (e.g. Linear classifiers, Support Vector Machines). KB features may not encompass subtler symptoms of the disease, nor be general enough to cover varying levels of severity of the disease. To overcome this limitation, some works have instead focused on speaker representation models, such as Gaussian Supervectors and i-vectors. For instance, Garcia et al. BIBREF1 proposed the use of i-vectors for PD classification and Laaridh et al. BIBREF6 applied the i-vector paradigm to the automatic prediction of several dysarthric speech evaluation metrics like intelligibility, severity, and articulation impairment. The intuition behind the use of these representations is the fact that these algorithms model speaker variability, which should include disease symptoms BIBREF1.
Proposed by Snyder et al., x-vectors are discriminative deep neural network-based speaker embeddings, that have outperformed i-vectors in tasks such as speaker and language recognition BIBREF7, BIBREF8, BIBREF9. Even though it may not be evident that discriminative data representations are suitable for disease detection when trained with general datasets (that do not necessarily include diseased patients), recent works have shown otherwise. X-vectors have been successfully applied to paralinguistic tasks such as emotion recognition BIBREF10, age and gender classification BIBREF11, the detection of obstructive sleep apnea BIBREF12 and as a complement to the detection of Alzheimer's Disease BIBREF0. Following this line of research, in this work we study the hypothesis that speaker characteristics embedded in x-vectors extracted from a single network, trained for speaker identification using general data, contain sufficient information to allow the detection of multiple diseases. Moreover, we aim to assess if this information is kept even when language mismatch is present, as has already been shown to be true for speaker recognition BIBREF8. In particular, we use the x-vector model as a feature extractor, to train Support Vector Machines for the detection of two speech-affecting diseases: Parkinson's disease (PD) and obstructive sleep apnea (OSA).
PD is the second most common neurodegenerative disorder of mid-to-late life after Alzheimer’s disease BIBREF13, affecting 1% of people over the age of 65. Common symptoms include bradykinesia (slowness or difficulty to perform movements), muscular rigidity, rest tremor, as well as postural and gait impairment. 89% of PD patients develop also speech disorders, typically hypokinetic dysarthria, which translates into symptoms such as reduced loudness, monoloudness, monopitch, hypotonicity, breathy and hoarse voice quality, and imprecise articulation BIBREF14BIBREF15.
OSA is a sleep-concerned breathing disorder characterized by a complete stop or decrease of the airflow, despite continued or increased inspiratory efforts BIBREF16. This disorder has a prevalence that ranges from 9% to 38% through different populations BIBREF17, with higher incidence in male and elderly groups. OSA causes mood and personality changes, depression, cognitive impairment, excessive daytime sleepiness, thus reducing the patients' quality of life BIBREF18, BIBREF19. It is also associated with diabetes, hypertension and cardiovascular diseases BIBREF16, BIBREF20. Moreover, undiagnosed sleep apnea can have a serious economic impact, having had an estimated cost of $\$150$ billion in the U.S, in 2015 BIBREF21. Considering the prevalence and serious nature of the two diseases described above, speech-based technology that tests for their existence has the potential to become a key tool for early detection, monitoring and prevention of these conditions BIBREF22.
The remainder of this document is organized as follows. Section SECREF2 presents the background concepts on speaker embeddings, and in particular on x-vectors. Section SECREF3 introduces the experimental setup: the corpora, the tasks, the KB features and the speaker embeddings employed. The results are presented and discussed in section SECREF4. Finally, section SECREF5 summarizes the main conclusions and suggests possible directions for future work.
Moreover, we aim to assess if this information is kept even when language mismatch is present, as has already been shown to be true for speaker recognition.
|
What do they aim to assess?
|
If this information is kept even when language mismatch is present.
|
null | false
| null |
Tell me whether or not these states are considered mountainous: Idaho, Vermont, Kansas, Hawaii, Florida, New Mexico, California, Rhode Island
|
Idaho, Vermont, Hawaii, New Mexico, and California are considered mountainous. Kansas Florida, and Rhode Island are not.
|
|
null | false
| null |
Why does it rain?
|
Rain is droplet of water falling from the sky. In the water life cycle, water from water bodies on earth evaporates and become clouds. The water droplets in the cloud coalesce and get bigger. When clouds become saturated, or get bigger and bigger, that the air can no longer hold it, they fall down as rain.
|
|
null | false
| null |
What is the best way to tune a snare drum?
|
The best way to tune a drum depends on which system the drum uses for tension. Newer technologies such as Dialtune's cable tuning system allow a drummer to quickly and evenly tune a drum head by turning separate dials for the top and bottom head.
Traditional lug and tension rod drums use a drum key and require more attention to detail. For these drums, start by getting all tension rods finger-tight. Bring the snare drum up to tension with a full turn of a drum key of each one, moving in a star pattern around the drum. Use a drumstick to tap around the drum by each tension rod and listen for a consistent sound.
Once you have the overall sound in the range you’re looking for, test the pitch by tapping around the drum. For the snare’s resonant head, detune it like the batter head. Tune it ‘tabletop tight’: so flat you could confidently rest a drink on it (don’t do this, though). Bring up each tension rod another quarter turn. You want to hear the wires, but not so much that it overwhelms the attack. Tap the head beside each tension rod, then start to even out the pitch with the tension rods so they all match.
|
|
null | false
| null |
What happens at a track meet?
|
Coaches, athletes, fans, and volunteers gather around a flat oval surface that is usually 160 to 400 meters in length. A track meet is a competition held on a day where people compete to win or improve their abilities as measured against a clock or measuring tape. Events at a track meet are categorized as running and field where athletes compete head to head against one another. Field events might consist of throwing a heavy ball, called shot put or jumping from a line, called long jump. Running events range from 50meters to many kilometers done by running laps around the oval. Usually athletes compete in heats or waves of 4-8 runners with the fastest overall winning the event. Families and friends gather to cheer on their athletes while coaches measure their performance while also giving advice during the track meet.
|
|
null | false
| null |
Samuel Fain Carter, the founder of Lumberman's Bank in Houston, commissioned the architecture firm of Sanguinet and Staats to design a sixteen-floor, steel-framed building on Main Street at the corner of Rusk Street in Houston. The Fort Worth-based Sanguinet and Staats had already been building skyscrapers in various cities in Texas, and was building a reputation for this type of structure.In 1909, the building had an estimated cost of $650,000. Carter planned to finance construction through issues of equity and debt, stipulating that he would restrict bonds to $400,000 in value. The Rice Institute agreed to purchase up to $200,000 in bonds.
|
Given this paragraph about the JW Marriott building in Downtown Houston, what was the framing made of and how tall was it?
|
The building used steel framing and was 16 floors.
|
null | false
| null |
John Johnstone (1881-1935) was a British businessman and rider. He was the tai-pan of the Jardine, Matheson & Co., member of the Shanghai Municipal Council and Legislative Council of Hong Kong.
He was the head of the Jardine, Matheson & Co. as well as director of numerous companies. He was member of the Shanghai Municipal Council and the Legislative Council of Hong Kong.
He was a keen rider and bore the reputation of being the best rider East of the Suez. He succeeded in winning the Jockey Cup on a pony named Ben-y-Gloe belonging to C. H. Ross, the former tai-pan of Jardines when he first went to Hong Kong in 1903. In 1904, he won the Professional Cup on a pony named Runaway Girl, purchased from W. A. Cruickshank. From 1903 to 1919, he rode in 1,178 races, winning 334, securing 192 seconds and 186 thirds, being unplaced 466 times, excluding the record of minor events such as Off Days or Gymkhanas. He won 14 Grand Nationals in China, and at one meeting in Hong Kong rode in 30 races winning 16 of them. He was also the President of the St. Andrew's Society.
He left Hong Kong on 31 March 1921 for home by the Empress of Asia.
In 1928 he won The Foxhunters Cup at Cheltenham riding Rathpatrick and officiated as steward at several of the northern meetings. He died at the Cheltenham meeting in 1935 where he was officiating.
|
What event John Johnstone won as a rider?
|
John Johnstone won the Jockey Cup on a horse named Ben-y-Gloe.
|
null | false
| null |
What is a bank?
|
A bank is a financial institution that accepts deposits from the public and creates a demand deposit while simultaneously making loans. Lending activities can be directly performed by the bank or indirectly through capital markets.
Because banks play an important role in financial stability and the economy of a country, most jurisdictions exercise a high degree of regulation over banks. Most countries have institutionalized a system known as fractional-reserve banking, under which banks hold liquid assets equal to only a portion of their current liabilities. In addition to other regulations intended to ensure liquidity, banks are generally subject to minimum capital requirements based on an international set of capital standards, the Basel Accords.
|
|
null | false
| null |
What is the English Premier League?
|
The English Premier League is the top flight of men’s football in England. The 38 match season extends from August to May, during which each of 20 teams play a home and away match against each of the other teams. A win earns a team 3 points, a draw is 1 point, and a loss is 0 points. The team with the most points at the end of the season wins. Unlike American sports, there are no playoffs to determine a champion. The 3 teams with the least points are relegated to the second level of English football, The Championship. To fill those spots in the next season, the top 3 teams from The Championship are promoted to the Premier League.
|
|
null | false
| null |
give me a list of all the things I can do with a piece of land.
|
- you can build a house
- you can build a farm
- you can graze cattle
- you can excavate for precious metals
- you can dig a well
- you can grow flowers
- you can grow an orchard
- you can build a park
- you can build a bunker under ground
|
|
null | false
| 384
|
Language model pre-training has proven to be highly effective in learning universal language representations from large-scale unlabeled data. ELMo BIBREF0, GPT BIBREF1 and BERT BIBREF2 have achieved great success in many NLP tasks, such as sentiment classification BIBREF3, natural language inference BIBREF4, and question answering BIBREF5.
Despite its empirical success, BERT's computational efficiency is a widely recognized issue because of its large number of parameters. For example, the original BERT-Base model has 12 layers and 110 million parameters. Training from scratch typically takes four days on 4 to 16 Cloud TPUs. Even fine-tuning the pre-trained model with task-specific dataset may take several hours to finish one epoch. Thus, reducing computational costs for such models is crucial for their application in practice, where computational resources are limited.
Motivated by this, we investigate the redundancy issue of learned parameters in large-scale pre-trained models, and propose a new model compression approach, Patient Knowledge Distillation (Patient-KD), to compress original teacher (e.g., BERT) into a lightweight student model without performance sacrifice. In our approach, the teacher model outputs probability logits and predicts labels for the training samples (extendable to additional unannotated samples), and the student model learns from the teacher network to mimic the teacher's prediction.
Different from previous knowledge distillation methods BIBREF6, BIBREF7, BIBREF8, we adopt a patient learning mechanism: instead of learning parameters from only the last layer of the teacher, we encourage the student model to extract knowledge also from previous layers of the teacher network. We call this `Patient Knowledge Distillation'. This patient learner has the advantage of distilling rich information through the deep structure of the teacher network for multi-layer knowledge distillation.
We also propose two different strategies for the distillation process: ($i$) PKD-Last: the student learns from the last $k$ layers of the teacher, under the assumption that the top layers of the original network contain the most informative knowledge to teach the student; and ($ii$) PKD-Skip: the student learns from every $k$ layers of the teacher, suggesting that the lower layers of the teacher network also contain important information and should be passed along for incremental distillation.
We evaluate the proposed approach on several NLP tasks, including Sentiment Classification, Paraphrase Similarity Matching, Natural Language Inference, and Machine Reading Comprehension. Experiments on seven datasets across these four tasks demonstrate that the proposed Patient-KD approach achieves superior performance and better generalization than standard knowledge distillation methods BIBREF6, with significant gain in training efficiency and storage reduction while maintaining comparable model accuracy to original large models. To the authors' best knowledge, this is the first known effort for BERT model compression.
To the authors' best knowledge, this is the first known effort for BERT model compression.
|
Is it the first known effort for BERT model compression?
|
Yes.
|
null | false
| null |
what is elasticsearch?
|
Elasticsearch is a search engine based on the Lucene library. It provides a distributed, multitenant-capable full-text search engine with an HTTP web interface and schema-free JSON documents.
|
|
null | false
| null |
Classify these as having wheels or not having wheels: car, plane, boat, bicycle, drone, street, motorcycle
|
Cars, planes, bicycles, and motorcycles typically have wheels. Boats, drones, and streets do not have wheels.
|
|
null | false
| null |
Tell me a story
|
Everyone carries an invisible bucket.
When the bucket is full then we are happy. When it is empty then we are sad.
The interesting thing is that we can only fill the bucket by doing good deeds and helping others. When I do a good deed to someone, my bucket gets filled but so does the other person's bucket.
When I do a bad deed to someone, then my bucket empties and also the other person's.
So, as long as we all keep filling each others buckets, we can all be happy
|
|
null | false
| null |
Jukebox is the eleventh studio album by Australian pop vocal group Human Nature released in October 2014. It contains covers from the 1950s and 1960s, with one original song, "End of Days"
|
Does Jukebox contain many original songs?
|
No, it contains covers from the 1950s and 1960s, with only one original song.
|
null | false
| null |
Dogecoin (/ˈdoʊ(d)ʒkɔɪn/ DOHJ-koyn or DOHZH-koyn, Abbreviation: DOGE; sign: Ð) is a cryptocurrency created by software engineers Billy Markus and Jackson Palmer, who decided to create a payment system as a "joke", making fun of the wild speculation in cryptocurrencies at the time. It is considered both the first "meme coin", and, more specifically, the first "dog coin". Despite its satirical nature, some consider it a legitimate investment prospect. Dogecoin features the face of the Shiba Inu dog from the "doge" meme as its logo and namesake. It was introduced on December 6, 2013, and quickly developed its own online community, reaching a market capitalization of over $85 billion on May 5, 2021. As of 2021, it is the sleeve sponsor of Watford Football Club.
Dogecoin.com promotes the currency as the "fun and friendly Internet currency", referencing its origins as a "joke". Software engineers Billy Markus and Jackson Palmer launched the satirical cryptocurrency as a way to make fun of Bitcoin and the many other cryptocurrencies boasting grand plans to take over the world. With the help of Reddit, the site became an instant hit. Within two weeks, Dogecoin had established a dedicated blog and forum, and its market value has reached US$8 million, once jumping to become the seventh largest electronic currency in the world. Dogecoin is based on Scrypt algorithm, and the transaction process is more convenient than Bitcoin.[citation needed] Dogecoin takes only 1 minute to confirm, while BTC takes 10 minutes.
|
Based on the reference text, provide a one sentence overview of Dogecoin
|
Dogecoin is a cryptocurrency that features the face of a dog the was created as a joke to make fun of Bitcoin and other cryptocurrencies.
|
null | false
| null |
Who has been elected as the 15th President of India?
|
Ms.Draupadi Murmu
|
|
null | false
| null |
Tell me 3 ingredients found in common United States foods that are banned in other countries. Provide an answer in list form and include examples of: foods that the ingredient is found in, countries where the ingredient is banned, and side effects.
|
Many popular foods consumed in the United States are banned in other countries due to the ingredients linked to adverse side effects.
• BHT (butylated hydroxytoluene), a food preservative, is often found in chewing gum, preserved meats, and dehydrated foods. BHT is banned in the EU and Japan due to evidence of tumor growth and impaired blood clotting.
• Food dyes such as Yellow No. 5 (Tartrazine) and Red No. 40 (Allura Red) are commonly added to colorful foods such as breakfast cereals, chips, popcorn, and soft drinks. Yellow No. 5 and Red No. 40 have been linked with the inhibition of cell development and are banned in Austria, Finland, France, and the United Kingdom.
• BVO (Brominated Vegetable Oil), is an additive used to prevent the separation of citrus flavoring in soft drinks and other beverages. BVO includes bromine, an element found in flame retardants, and linked with neurologic symptoms including impaired balance, headaches, and memory loss.
|
|
null | false
| null |
The new UCLA campus in 1929 had four buildings: Royce Hall and Haines Hall on the north, and Powell Library and Kinsey Hall (now called Renee And David Kaplan Hall) on the south. The Janss Steps were the original 87-step entrance to the university that lead to the quad of these four buildings. Today, the campus includes 163 buildings across 419 acres (1.7 km2) in the western part of Los Angeles, north of the Westwood shopping district and just south of Sunset Boulevard. In terms of acreage, it is the second-smallest of the ten UC campuses. The campus is approximately 1 mile east of I-405 (the San Diego Freeway).
The campus is in the residential area of Westwood and bordered by Bel-Air to the north, Beverly Hills to the east, and Brentwood to the west. The campus is informally divided into North Campus and South Campus, which are both on the eastern half of the university's land. North Campus is the original campus core; its buildings are more traditional in appearance and clad in imported Italian brick. North Campus is home to the arts, humanities, social sciences, law, history, and business programs and is centered around ficus and sycamore-lined Dickson Court, also known as the "Sunken Garden". South Campus is home to the physical sciences, life sciences, engineering, mathematical sciences, health-related fields, and the UCLA Medical Center. The campus includes sculpture gardens, fountains, museums, and a mix of architectural styles.
Ackerman Union, the John Wooden Center, the Arthur Ashe Health and Wellness Center, the Student Activities Center, Kerckhoff Hall, the J.D. Morgan Center, the James West Alumni Center, and Pauley Pavilion stand at the center of the campus, bordering Wilson Plaza. The campus is bisected by Bruin Walk, a heavily traveled pathway from the residential hill to the main campus. At the intersection of Bruin Walk and Westwood Plaza is Bruin Plaza, featuring an outdoor performing arts stage and a bronze statue of the Bruin bear.
During the 2028 Summer Olympics, wrestling and judo will be held at Pauley Pavilion. The campus will also be the location of the Olympic Village for the Olympic and Paralympic Games.
In September 2022 UCLA purchased the Palos Verdes (PV) campus of Marymount California University, which had closed its doors that August, after fifty years of classes; the PV campus joins two other locations, in downtown LA, and in Culver City as satellites of the Westwood campus. A faculty and administration task force will review possible applications of the PV campus for UCLA. The Channel Islands will again be visible from a UCLA campus.
|
Based on this passage about UCLA, tell me the difference between North Campus and South Campus and which residential areas border the campus.
|
At UCLA, North Campus hosts the arts, humanities, social sciences, law, history, and business programs, while South Campus is home to the physical sciences, life sciences, engineering, mathematical sciences, health-related fields, and the UCLA Medical Center. The UCLA Campus is bordered by Bel Air, Brentwood, and Beverly Hills.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.