This paper introduces a novel framework that combines graph-driven context retrieval in conjunction to knowledge graphs based enhancement, honing the proficiency of LLMs, especially in domain specific community question answering platforms like AskUbuntu, Unix, and ServerFault. We conduct experiments on various LLMs with different parameter sizes to evaluate their ability to ground knowledge and determine factual accuracy in answers to open-ended questions. Our methodology GraphContextGen consistently outperforms dominant text-based retrieval systems, demonstrating its robustness and adaptability to a larger number of use cases. This advancement highlights the importance of pairing context rich data retrieval with LLMs, offering a renewed approach to knowledge sourcing and generation in AI systems. We also show that, due to rich contextual data retrieval, the crucial entities, along with the generated answer, remain factually coherent with the gold answer. We shall release the source code and datasets upon acceptance.

With the rise of online abuse, the NLP community has begun investigating the use of neural architectures to generate counterspeech that can “counter” the vicious tone of such abusive speech and dilute/ameliorate their rippling effect over the social network. However, most of the efforts so far have been primarily focused on English. To bridge the gap for low-resource languages such as Bengali and Hindi, we create a benchmark dataset of 5,062 abusive speech/counterspeech pairs, of which 2,460 pairs are in Bengali, and 2,602 pairs are in Hindi. We implement several baseline models considering various interlingual transfer mechanisms with different configurations to generate suitable counterspeech to set up an effective benchmark. We observe that the monolingual setup yields the best performance. Further, using synthetic transfer, language models can generate counterspeech to some extent; specifically, we notice that transferability is better when languages belong to the same language family.

Large Language Models (LLMs) exhibit impressive zero/few-shot inference and generation quality for high-resource languages (HRLs). A few of them have been trained on low-resource languages (LRLs) and give decent performance. Owing to the prohibitive costs of training LLMs, they are usually used as a network service, with the client charged by the count of input and output tokens. The number of tokens strongly depends on the script and language, as well as the LLM’s subword vocabulary. We show that LRLs are at a pricing disadvantage, because the well-known LLMs produce more tokens for LRLs than HRLs. This is because most currently popular LLMs are optimized for HRL vocabularies. Our objective is to level the playing field: reduce the cost of processing LRLs in contemporary LLMs while ensuring that predictive and generative qualities are not compromised. As means to reduce the number of tokens processed by the LLM, we consider code-mixing, translation, and transliteration of LRLs to HRLs. We perform an extensive study using the IndicXTREME classification and six generative tasks dataset, covering 15 Indic and 3 other languages, while using GPT-4 (one of the costliest LLM services released so far) as a commercial LLM. We observe and analyze interesting patterns involving token count, cost, and quality across a multitude of languages and tasks. We show that choosing the best policy to interact with the LLM can reduce cost by ~90% while giving better or comparable performance, compared to communicating with the LLM in the original LRL.

We investigate the knowledge of object affordances in pre-trained language models (LMs) and pre-trained Vision-Language models (VLMs).A growing body of literature shows that PTLMs fail inconsistently and non-intuitively, demonstrating a lack of reasoning and grounding. To take a first step toward quantifying the effect of grounding (or lack thereof), we curate a novel and comprehensive dataset of object affordances – Text2Afford, characterized by 15 affordance classes. Unlike affordance datasets collected in vision and language domains, we annotate in-the-wild sentences with objects and affordances. Experimental results reveal that PTLMs exhibit limited reasoning abilities when it comes to uncommon object affordances. We also observe that pre-trained VLMs do not necessarily capture object affordances effectively. Through few-shot fine-tuning, we demonstrate improvement in affordance knowledge in PTLMs and VLMs. Our research contributes a novel dataset for language grounding tasks, and presents insights into LM capabilities, advancing the understanding of object affordances.

Counterspeech presents a viable alternative to banning or suspending users for hate speech while upholding freedom of expression. However, writing effective counterspeech is challenging for moderators/users. Hence, developing suggestion tools for writing counterspeech is the need of the hour. One critical challenge in developing such a tool is the lack of quality and diversity of the responses in the existing datasets. Hence, we introduce a new dataset - CrowdCounter containing 3,425 hate speech-counterspeech pairs spanning six different counterspeech types (empathy, humor, questioning, warning, shaming, contradiction), which is the first of its kind. The design of our annotation platform itself encourages annotators to write type-specific, non-redundant and high-quality counterspeech. We evaluate two frameworks for generating counterspeech responses - vanilla and type-controlled prompts - across four large language models. In terms of metrics, we evaluate the responses using relevance, diversity and quality. We observe that Flan-T5 is the best model in the vanilla framework across different models. Type-specific prompts enhance the relevance of the responses, although they might reduce the language quality. DialoGPT proves to be the best at following the instructions and generating the type-specific counterspeech accurately.

Large language models like ChatGPT have recently shown a great promise in performing several tasks, including hate speech detection. However, it is crucial to comprehend the limitations of these models to build robust hate speech detection systems. To bridge this gap, our study aims to evaluate the strengths and weaknesses of the ChatGPT model in detecting hate speech at a granular level across 11 languages. Our evaluation employs a series of functionality tests that reveals various intricate failures of the model which the aggregate metrics like macro F1 or accuracy are not able to unfold. In addition, we investigate the influence of complex emotions, such as the use of emojis in hate speech, on the performance of the ChatGPT model. Our analysis highlights the shortcomings of the generative models in detecting certain types of hate speech and highlighting the need for further research and improvements in the workings of these models.

Recently, influence functions present an apparatus for achieving explainability for deep neural models by quantifying the perturbation of individual train instances that might impact a test prediction. Our objectives in this paper are twofold. First we incorporate influence functions as a feedback into the model to improve its performance. Second, in a dataset extension exercise, using influence functions to automatically identify data points that have been initially ‘silver’ annotated by some existing method and need to be cross-checked (and corrected) by annotators to improve the model performance. To meet these objectives, in this paper, we introduce InfFeed, which uses influence functions to compute the influential instances for a target instance. Toward the first objective, we adjust the label of the target instance based on its influencer(s) label. In doing this, InfFeed outperforms the state-of-the-art baselines (including LLMs) by a maximum macro F1-score margin of almost 4% for hate speech classification, 3.5% for stance classification, and 3% for irony and 2% for sarcasm detection. Toward the second objective we show that manually re-annotating only those silver annotated data points in the extension set that have a negative influence can immensely improve the model performance bringing it very close to the scenario where all the data points in the extension set have gold labels. This allows for huge reduction of the number of data points that need to be manually annotated since out of the silver annotated extension dataset, the influence function scheme picks up ~1/1000 points that need manual correction.

With the emergence of numerous Large Language Models (LLM), the usage of such models in various Natural Language Processing (NLP) applications is increasing extensively. Counterspeech generation is one such key task where efforts are made to develop generative models by fine-tuning LLMs with hatespeech - counterspeech pairs, but none of these attempts explores the intrinsic properties of large language models in zero-shot settings. In this work, we present a comprehensive analysis of the performances of four LLMs namely GPT-2, DialoGPT, ChatGPT and FlanT5 in zero-shot settings for counterspeech generation, which is the first of its kind. For GPT-2 and DialoGPT, we further investigate the deviation in performance with respect to the sizes (small, medium, large) of the models. On the other hand, we propose three different prompting strategies for generating different types of counterspeech and analyse the impact of such strategies on the performance of the models. Our analysis shows that there is an improvement in generation quality for two datasets (17%), however the toxicity increase (25%) with increase in model size. Considering type of model, GPT-2 and FlanT5 models are significantly better in terms of counterspeech quality but also have high toxicity as compared to DialoGPT. ChatGPT are much better at generating counter speech than other models across all metrics. In terms of prompting, we find that our proposed strategies help in improving counter speech generation across all the models.

Multilingual language models (MLLMs) like mBERTpromise to extend the benefits of NLP research to low-resource languages (LRLs). However, LRL words are under-represented in the wordpiece/subword vocabularies of MLLMs. This leads to many LRL words getting replaced by UNK, or concatenated from morphologically unrelated wordpieces, leading to low task accuracy. (Pre)-training MLLMs after including LRL documents is resource-intensive in terms of both human inputs and computational resources. In response, we propose EVALM (entropy-based vocabulary augmented language model), which uses a new task-cognizant measurement to detect the most vulnerable LRL words, whose wordpiece segmentations are undesirable. EVALM then provides reasonable initializations of their embeddings, followed by limited fine-tuning using the small LRL task corpus. Our experiments show significant performance improvements and also some surprising limits to such vocabulary augmentation strategies in various classification tasks for multiple diverse LRLs, as well as code-mixed texts. We will release the code and data to enable further research.

Recently efforts have been made by social media platforms as well as researchers to detect hateful or toxic language using large language models. However, none of these works aim to use explanation, additional context and victim community information in the detection process. We utilise different prompt variation, input information and evaluate large language models in zero shot setting (without adding any in-context examples). We select two large language models (GPT-3.5 and text-davinci) and three datasets - HateXplain, implicit hate and ToxicSpans. We find that on average including the target information in the pipeline improves the model performance substantially (∼20-30%) over the baseline across the datasets. There is also a considerable effect of adding the rationales/explanations into the pipeline (∼10-20%) over the baseline across the datasets. In addition, we further provide a typology of the error cases where these large language models fail to (i) classify and (ii) explain the reason for the decisions they take. Such vulnerable points automatically constitute ‘jailbreak’ prompts for these models and industry scale safeguard techniques need to be developed to make the models robust against such prompts.

The dramatic increase in the use of social media platforms for information sharing has also fueled a steep growth in online abuse. A simple yet effective way of abusing individuals or communities is by creating memes, which often integrate an image with a short piece of text layered on top of it. Such harmful elements are in rampant use and are a threat to online safety. Hence it is necessary to develop efficient models to detect and flag abusive memes. The problem becomes more challenging in a low-resource setting (e.g., Bengali memes, i.e., images with Bengali text embedded on it) because of the absence of benchmark datasets on which AI models could be trained. In this paper we bridge this gap by building a Bengali meme dataset. To setup an effective benchmark we implement several baseline models for classifying abusive memes using this dataset. We observe that multimodal models that use both textual and visual information outperform unimodal models. Our best-performing model achieves a macro F1 score of 70.51. Finally, we perform a qualitative error analysis of the misclassified memes of the best-performing text-based, image-based and multimodal models.

The last decade has witnessed a surge in the interaction of people through social networking platforms. While there are several positive aspects of these social platforms, their proliferation has led them to become the breeding ground for cyber-bullying and hate speech. Recent advances in NLP have often been used to mitigate the spread of such hateful content. Since the task of hate speech detection is usually applicable in the context of social networks, we introduce CRUSH, a framework for hate speech detection using User Anchored self-supervision and contextual regularization. Our proposed approach secures ~1-12% improvement in test set metrics over best performing previous approaches on two types of tasks and multiple popular English language social networking datasets.

Due to the sheer volume of online hate, the AI and NLP communities have started building models to detect such hateful content. Recently, multilingual hate is a major emerging challenge for automated detection where code-mixing or more than one language have been used for conversation in social media. Typically, hate speech detection models are evaluated by measuring their performance on the held-out test data using metrics such as accuracy and F1-score. While these metrics are useful, it becomes difficult to identify using them where the model is failing, and how to resolve it. To enable more targeted diagnostic insights of such multilingual hate speech models, we introduce a set of functionalities for the purpose of evaluation. We have been inspired to design this kind of functionalities based on real-world conversation on social media. Considering Hindi as a base language, we craft test cases for each functionality. We name our evaluation dataset HateCheckHIn. To illustrate the utility of these functionalities , we test state-of-the-art transformer based m-BERT model and the Perspective API.

Social media often serves as a breeding ground for various hateful and offensive content. Identifying such content on social media is crucial due to its impact on the race, gender, or religion in an unprejudiced society. However, while there is extensive research in hate speech detection in English, there is a gap in hateful content detection in low-resource languages like Bengali. Besides, a current trend on social media is the use of Romanized Bengali for regular interactions. To overcome the existing research’s limitations, in this study, we develop an annotated dataset of 10K Bengali posts consisting of 5K actual and 5K Romanized Bengali tweets. We implement several baseline models for the classification of such hateful posts. We further explore the interlingual transfer mechanism to boost classification performance. Finally, we perform an in-depth error analysis by looking into the misclassified posts by the models. While training actual and Romanized datasets separately, we observe that XLM-Roberta performs the best. Further, we witness that on joint training and few-shot training, MuRIL outperforms other models by interpreting the semantic expressions better. We make our code and dataset public for others.

Social media platforms often act as breeding grounds for various forms of trolling or malicious content targeting users or communities. One way of trolling users is by creating memes, which in most cases unites an image with a short piece of text embedded on top of it. The situation is more complex for multilingual(e.g., Tamil) memes due to the lack of benchmark datasets and models. We explore several models to detect Troll memes in Tamil based on the shared task, “Troll Meme Classification in DravidianLangTech2022” at ACL-2022. We observe while the text-based model MURIL performs better for Non-troll meme classification, the image-based model VGG16 performs better for Troll-meme classification. Further fusing these two modalities help us achieve stable outcomes in both classes. Our fusion model achieved a 0.561 weighted average F1 score and ranked second in this task.

Relation classification (sometimes called ‘extraction’) requires trustworthy datasets for fine-tuning large language models, as well as for evaluation. Data collection is challenging for Indian languages, because they are syntactically and morphologically diverse, as well as different from resource-rich languages like English. Despite recent interest in deep generative models for Indian languages, relation classification is still not well-served by public data sets. In response, we present IndoRE, a dataset with 39K entity- and relation-tagged gold sentences in three Indian languages, plus English. We start with a multilingual BERT (mBERT) based system that captures entity span positions and type information and provides competitive monolingual relation classification. Using this system, we explore and compare transfer mechanisms between languages. In particular, we study the accuracy-efficiency tradeoff between expensive gold instances vs. translated and aligned ‘silver’ instances.

Social media often acts as breeding grounds for different forms of offensive content. For low resource languages like Tamil, the situation is more complex due to the poor performance of multilingual or language-specific models and lack of proper benchmark datasets. Based on this shared task “Offensive Language Identification in Dravidian Languages” at EACL 2021; we present an exhaustive exploration of different transformer models, We also provide a genetic algorithm technique for ensembling different models. Our ensembled models trained separately for each language secured the first position in Tamil, the second position in Kannada, and the first position in Malayalam sub-tasks. The models and codes are provided.

In this paper, we demonstrate how code-switching patterns can be utilised to improve various downstream NLP applications. In particular, we encode various switching features to improve humour, sarcasm and hate speech detection tasks. We believe that this simple linguistic observation can also be potentially helpful in improving other similar NLP applications.

Millions of people irrespective of socioeconomic and demographic backgrounds, depend on Wikipedia articles everyday for keeping themselves informed regarding popular as well as obscure topics. Articles have been categorized by editors into several quality classes, which indicate their reliability as encyclopedic content. This manual designation is an onerous task because it necessitates profound knowledge about encyclopedic language, as well navigating circuitous set of wiki guidelines. In this paper we propose Neural wikipedia Quality Monitor (NwQM), a novel deep learning model which accumulates signals from several key information sources such as article text, meta data and images to obtain improved Wikipedia article representation. We present comparison of our approach against a plethora of available solutions and show 8% improvement over state-of-the-art approaches with detailed ablation studies.

This paper describes our approach to solve Semeval task 3: EmoContext; where, given a textual dialogue i.e. a user utterance along with two turns of context, we have to classify the emotion associated with the utterance as one of the following emotion classes: Happy, Sad, Angry or Others. To solve this problem, we experiment with different deep learning models ranging from simple bidirectional LSTM (Long and short term memory) model to comparatively complex attention model. We also experiment with word embedding conceptnet along with word embedding generated from bi-directional LSTM taking input characters. We fine-tune different parameters and hyper-parameters associated with each of our models and report the value of evaluating measure i.e. micro precision along with class wise precision, recall and F1-score of each system. We report the bidirectional LSTM model, along with the input word embedding as the concatenation of word embedding generated from bidirectional LSTM for word characters and conceptnet embedding, as the best performing model with a highest micro-F1 score of 0.7261. We also report class wise precision, recall, and f1-score of best performing model along with other models that we have experimented with.

The compositionality degree of multiword expressions indicates to what extent the meaning of a phrase can be derived from the meaning of its constituents and their grammatical relations. Prediction of (non)-compositionality is a task that has been frequently addressed with distributional semantic models. We introduce a novel technique to blend hierarchical information with distributional information for predicting compositionality. In particular, we use hypernymy information of the multiword and its constituents encoded in the form of the recently introduced Poincaré embeddings in addition to the distributional information to detect compositionality for noun phrases. Using a weighted average of the distributional similarity and a Poincaré similarity function, we obtain consistent and substantial, statistically significant improvement across three gold standard datasets over state-of-the-art models based on distributional information only. Unlike traditional approaches that solely use an unsupervised setting, we have also framed the problem as a supervised task, obtaining comparable improvements. Further, we publicly release our Poincaré embeddings, which are trained on the output of handcrafted lexical-syntactic patterns on a large corpus.

Wikipedia can easily be justified as a behemoth, considering the sheer volume of content that is added or removed every minute to its several projects. This creates an immense scope, in the field of natural language processing toward developing automated tools for content moderation and review. In this paper we propose Self Attentive Revision Encoder (StRE) which leverages orthographic similarity of lexical units toward predicting the quality of new edits. In contrast to existing propositions which primarily employ features like page reputation, editor activity or rule based heuristics, we utilize the textual content of the edits which, we believe contains superior signatures of their quality. More specifically, we deploy deep encoders to generate representations of the edits from its text content, which we then leverage to infer quality. We further contribute a novel dataset containing ∼ 21M revisions across 32K Wikipedia pages and demonstrate that StRE outperforms existing methods by a significant margin – at least 17% and at most 103%. Our pre-trained model achieves such result after retraining on a set as small as 20% of the edits in a wikipage. This, to the best of our knowledge, is also the first attempt towards employing deep language models to the enormous domain of automated content moderation and review in Wikipedia.

In this paper we propose a deep learning framework for sarcasm target detection in predefined sarcastic texts. Identification of sarcasm targets can help in many core natural language processing tasks such as aspect based sentiment analysis, opinion mining etc. To begin with, we perform an empirical study of the socio-linguistic features and identify those that are statistically significant in indicating sarcasm targets (p-values in the range(0.05,0.001)). Finally, we present a deep-learning framework augmented with socio-linguistic features to detect sarcasm targets in sarcastic book-snippets and tweets. We achieve a huge improvement in the performance in terms of exact match and dice scores compared to the current state-of-the-art baseline.

We present CL Scholar, the ACL Anthology knowledge graph miner to facilitate high-quality search and exploration of current research progress in the computational linguistics community. In contrast to previous works, periodically crawling, indexing and processing of new incoming articles is completely automated in the current system. CL Scholar utilizes both textual and network information for knowledge graph construction. As an additional novel initiative, CL Scholar supports more than 1200 scholarly natural language queries along with standard keyword-based search on constructed knowledge graph. It answers binary, statistical and list based natural language queries. The current system is deployed at http://cnerg.iitkgp.ac.in/aclakg. We also provide REST API support along with bulk download facility. Our code and data are available at https://github.com/CLScholar.

The exponential increase in the usage of Wikipedia as a key source of scientific knowledge among the researchers is making it absolutely necessary to metamorphose this knowledge repository into an integral and self-contained source of information for direct utilization. Unfortunately, the references which support the content of each Wikipedia entity page, are far from complete. Why are the reference section ill-formed for most Wikipedia pages? Is this section edited as frequently as the other sections of a page? Can there be appropriate surrogates that can automatically enhance the reference section? In this paper, we propose a novel two step approach – WikiRef – that (i) leverages the wikilinks present in a scientific Wikipedia target page and, thereby, (ii) recommends highly relevant references to be included in that target page appropriately and automatically borrowed from the reference section of the wikilinks. In the first step, we build a classifier to ascertain whether a wikilink is a potential source of reference or not. In the following step, we recommend references to the target page from the reference section of the wikilinks that are classified as potential sources of references in the first step. We perform an extensive evaluation of our approach on datasets from two different domains – Computer Science and Physics. For Computer Science we achieve a notably good performance with a precision@1 of 0.44 for reference recommendation as opposed to 0.38 obtained from the most competitive baseline. For the Physics dataset, we obtain a similar performance boost of 10% with respect to the most competitive baseline.

This paper describes the systems developed for 1st and 2nd tasks of the 3rd Social Media Mining for Health Applications Shared Task at EMNLP 2018. The first task focuses on automatic detection of posts mentioning a drug name or dietary supplement, a binary classification. The second task is about distinguishing the tweets that present personal medication intake, possible medication intake and non-intake. We performed extensive experiments with various classifiers like Logistic Regression, Random Forest, SVMs, Gradient Boosted Decision Trees (GBDT) and deep learning architectures such as Long Short-Term Memory Networks (LSTM), jointed Convolutional Neural Networks (CNN) and LSTM architecture, and attention based LSTM architecture both at word and character level. We have also explored using various pre-trained embeddings like Global Vectors for Word Representation (GloVe), Word2Vec and task-specific embeddings learned using CNN-LSTM and LSTMs.

n this paper, we present a set of computational methods to identify the likeliness of a word being borrowed, based on the signals from social media. In terms of Spearman’s correlation values, our methods perform more than two times better (∼ 0.62) in predicting the borrowing likeliness compared to the best performing baseline (∼ 0.26) reported in literature. Based on this likeliness estimate we asked annotators to re-annotate the language tags of foreign words in predominantly native contexts. In 88% of cases the annotators felt that the foreign language tag should be replaced by native language tag, thus indicating a huge scope for improvement of automatic language identification systems.

Word senses are not static and may have temporal, spatial or corpus-specific scopes. Identifying such scopes might benefit the existing WSD systems largely. In this paper, while studying corpus specific word senses, we adapt three existing predominant and novel-sense discovery algorithms to identify these corpus-specific senses. We make use of text data available in the form of millions of digitized books and newspaper archives as two different sources of corpora and propose automated methods to identify corpus-specific word senses at various time points. We conduct an extensive and thorough human judgement experiment to rigorously evaluate and compare the performance of these approaches. Post adaptation, the output of the three algorithms are in the same format and the accuracy results are also comparable, with roughly 45-60% of the reported corpus-specific senses being judged as genuine.

This paper proposes OCR++, an open-source framework designed for a variety of information extraction tasks from scholarly articles including metadata (title, author names, affiliation and e-mail), structure (section headings and body text, table and figure headings, URLs and footnotes) and bibliography (citation instances and references). We analyze a diverse set of scientific articles written in English to understand generic writing patterns and formulate rules to develop this hybrid framework. Extensive evaluations show that the proposed framework outperforms the existing state-of-the-art tools by a large margin in structural information extraction along with improved performance in metadata and bibliography extraction tasks, both in terms of accuracy (around 50% improvement) and processing time (around 52% improvement). A user experience study conducted with the help of 30 researchers reveals that the researchers found this system to be very helpful. As an additional objective, we discuss two novel use cases including automatically extracting links to public datasets from the proceedings, which would further accelerate the advancement in digital libraries. The result of the framework can be exported as a whole into structured TEI-encoded documents. Our framework is accessible online at http://www.cnergres.iitkgp.ac.in/OCR++/home/.

We present a study of the word interaction networks of Bengali in the framework of complex networks. The topological properties of these networks reveal interesting insights into the morpho-syntax of the language, whereas clustering helps in the induction of the natural word classes leading to a principled way of designing POS tagsets. We compare different network construction techniques and clustering algorithms based on the cohesiveness of the word clusters. Cohesiveness is measured against two gold-standard tagsets by means of the novel metric of tag-entropy. The approach presented here is a generic one that can be easily extended to any language.