Abstract

Prior work on constructing challenging tabular inference data centered primarily on human annotation or automatic synthetic generation. Both techniques have their own set of issues. Human annotation, despite its diversity and superior reasoning, struggles from scaling concerns. Synthetic data, on the other hand, despite its scalability, suffers from lack of linguistic and reasoning diversity. In this paper, we address both of these concerns by presenting a recasting approach that semi-automatically generates tabular NLI instances. We transform the table2text dataset ToTTo (Parikh et al., 2020) into a tabular NLI dataset using our proposed framework. We demonstrate the use of our recasted data as an evaluation benchmark as well as augmentation data to improve performance on TabFact (Chen et al., 2020b). Furthermore, we test the effectiveness of models trained on our data on the TabFact benchmark in the zeroshot scenario.

Abstract

Recent state of the art models and new datasets have advanced many Natural Language Processing areas, especially, Machine Reading Comprehension tasks have improved with the help of datasets like SQuAD (Stanford Question Answering Dataset). But, large high quality datasets are still not a reality for low resource languages like Telugu to record progress in MRC. In this paper, we present a Telugu Question Answering Dataset - TeQuAD with the size of 82k parallel triples created by translating triples from the SQuAD. We also introduce a few methods to create similar Question Answering datasets for the low resource languages. Then, we present the performance of our models which outperform baseline models on Monolingual and Cross Lingual Machine Reading Comprehension (CLMRC) setups, the best of them resulting in an F1 score of 83 % and Exact Match (EM) score of 61 %

Abstract

Automatic extraction of “significant” components of a legal contract, has the potential to simplify the end user’s comprehension. In essence, “significant” pieces of information have 1) information pertaining to material/practical details about a specific contract and 2) information that is novel or comes as a “surprise” for a specific type of contract. It indicates that the significance of a component may be defined at an individual contract level and at a contract-type level. A component, sentence, or paragraph, may be considered significant at a contract level if it contains contract-specific information (CSI), like names, dates, or currency terms. At a contract-type level, components that deviate significantly from the norm for the type may be considered significant (type-specific information (TSI)). In this paper, we present approaches to extract “significant” components from a contract at both these levels. We attempt to do this by identifying patterns in a pool of documents of the same kind. Consequently, in our approach, the solution is formulated in two parts: identifying CSI using a BERT-based contract-specific information extractor and identifying TSI by scoring sentences in a contract for their likelihood. In this paper, we even describe the annotated corpus of contract documents that we created as a first step toward the development of such a language-processing system. We also release a dataset of contract samples containing sentences belonging to CSI and TSI.

Abstract

Code mixing is the linguistic phenomenon where bilingual speakers tend to switch between two or more languages in conversations. Recent work on code-mixing in computational settings has leveraged social media code mixed texts to train NLP models. For capturing the variety of code mixing in, and across corpus, Language ID (LID) tags based measures (CMI) have been proposed. Syntactical variety/patterns of code-mixing and their relationship vis-a-vis computational model’s performance is under explored. In this work, we investigate a collection of English(en)-Hindi(hi) code-mixed datasets from a syntactic lens to propose, SyMCoM, an indicator of syntactic variety in code-mixed text, with intuitive theoretical bounds. We train SoTA en-hi PoS tagger, accuracy of 93.4%, to reliably compute PoS tags on a corpus, and demonstrate the utility of SyMCoM by applying it on various syntactical categories on a collection of datasets, and compare datasets using the measure.

Abstract

Hashtag segmentation is the task of breaking a hashtag into its constituent tokens. Hashtags often encode the essence of usergenerated posts, along with information like topic and sentiment, which are useful in downstream tasks. Hashtags prioritize brevity and are written in unique ways - transliterating and mixing languages, spelling variations, creative named entities. Benchmark datasets used for the hashtag segmentation task - STAN, BOUN - are small in size and extracted from a single set of tweets. However, datasets should reflect the variations in writing styles of hashtags and also account for domain and language specificity, failing which the results will misrepresent model performance. We argue that model performance should be assessed on a wider variety of hashtags, and datasets should be carefully curated. To this end, we propose HashSet, a dataset comprising of: a) 1.9k manually annotated dataset; b) 3.3M loosely supervised dataset. HashSet dataset is sampled from a different set of tweets when compared to existing datasets and provides an alternate distribution of hashtags to build and validate hashtag segmentation models. We show that the performance of SOTA models for Hashtag Segmentation drops substantially on proposed dataset, indicating that the proposed dataset provides an alternate set of hashtags to train and assess models. Datasets and results are released publicly and can be accessed from https://github.com/prashantkodali/HashSet

Abstract

Often questions provided to open-domain question answering systems are ambiguous. Traditional QA systems that provide a single answer are incapable of answering ambiguous questions since the question may be interpreted in several ways and may have multiple distinct answers. In this paper, we address multi-answer retrieval which entails retrieving passages that can capture majority of the diverse answers to the question. We propose a re-ranking based approach using Determinantal point processes utilizing BERT as kernels. Our method jointly considers query-passage relevance and passage-passage correlation to retrieve passages that are both query-relevant and diverse. Results demonstrate that our re-ranking technique outperforms state-of-the-art method on the AmbigQA dataset.

Abstract

Named Entity Recognition (NER) is a critical task in the field of Natural Language Processing (NLP) and is also a sub-task of Information Extraction. There has been a significant amount of work done in entity extraction and Named Entity Recognition for resource-rich languages. Entity extraction from code-mixed social media data like tweets from twitter complicates the problem due to its unstructured, informal, and incomplete information available in tweets. Here, we present work on NER in Kannada-English code-mixed social media corpus with corresponding named entity tags referring to Organisation (Org), Person (Pers), and Location (Loc). We experimented with machine learning classification models like Conditional Random Fields (CRF), Bi-LSTM, and Bi-LSTM-CRF models on our corpus.

Abstract

Automatic extraction of “significant” components of a legal contract, has the potential to simplify the end user’s comprehension. In essence, “significant” pieces of information have 1) information pertaining to material/practical details about a specific contract and 2) information that is novel or comes as a “surprise” for a specific type of contract. It indicates that the significance of a component may be defined at an individual contract level and at a contract-type level. A component, sentence, or paragraph, may be considered significant at a contract level if it contains contract-specific information (CSI), like names, dates, or currency terms. At a contract-type level, components that deviate significantly from the norm for the type may be considered significant (type-specific information (TSI)). In this paper, we present approaches to extract “significant” components from a contract at both these levels. We attempt to do this by identifying patterns in a pool of documents of the same kind. Consequently, in our approach, the solution is formulated in two parts: identifying CSI using a BERT-based contract-specific information extractor and identifying TSI by scoring sentences in a contract for their likelihood. In this paper, we even describe the annotated corpus of contract documents that we created as a first step toward the development of such a language-processing system. We also release a dataset of contract samples containing sentences belonging to CSI and TSI.

Abstract

Causality detection and identification is centered on identifying semantic and cognitive connections in a sentence. In this paper, we describe the effort of team LTRC for Causal News Corpus - Event Causality Shared Task 2022 at the 5th Workshop on Challenges and Applications of Automated Extraction of Sociopolitical Events from Text (CASE 2022) (Tan et al., 2022a). The shared task consisted of two subtasks: 1) identifying if a sentence contains a causality relation, and 2) identifying spans of text that correspond to cause, effect and signals. We fine-tuned transformer-based models with adapters for both subtasks. Our best-performing models obtained a binary F1 score of 0.853 on held-out data for subtask 1 and a macro F1 score of 0.032 on held-out data for subtask 2. Our approach is ranked third in subtask 1 and fourth in subtask 2. The paper describes our experiments, solutions, and analysis in detail.

Abstract

Aspect extraction is not a well-explored topic in Hindi, with only one corpus having been developed for the task. In this paper, we dis- cuss the merits of the existing corpus in terms of quality, size, sparsity, and performance in aspect extraction tasks using established mod- els. To provide a better baseline corpus for aspect extraction, we translate the SemEval 2014 aspect-based sentiment analysis dataset and annotate the aspects in that data. We provide rigorous guidelines and a replicable methodology for this task. We quantitatively evaluate the translations and annotations us- ing inter-annotator agreement scores. We also evaluate our dataset using state-of-the-art neu- ral aspect extraction models in both monolin- gual and multilingual settings and show that the models perform far better on our corpus than on the existing Hindi dataset. With this, we establish our corpus as the gold-standard aspect extraction dataset in Hindi