Kripabandhu Ghosh

Abhay Shukla, Paheli Bhattacharya, Soham Poddar et al.

Summarization of legal case judgement documents is a challenging problem in Legal NLP. However, not much analyses exist on how different families of summarization models (e.g., extractive vs. abstractive) perform when applied to legal case documents. This question is particularly important since many recent transformer-based abstractive summarization models have restrictions on the number of input tokens, and legal documents are known to be very long. Also, it is an open question on how best to evaluate legal case document summarization systems. In this paper, we carry out extensive experiments with several extractive and abstractive summarization methods (both supervised and unsupervised) over three legal summarization datasets that we have developed. Our analyses, that includes evaluation by law practitioners, lead to several interesting insights on legal summarization in specific and long document summarization in general.

Aniket Deroy, Kripabandhu Ghosh, Saptarshi Ghosh

Automatic summarization of legal case judgements has traditionally been attempted by using extractive summarization methods. However, in recent years, abstractive summarization models are gaining popularity since they can generate more natural and coherent summaries. Legal domain-specific pre-trained abstractive summarization models are now available. Moreover, general-domain pre-trained Large Language Models (LLMs), such as ChatGPT, are known to generate high-quality text and have the capacity for text summarization. Hence it is natural to ask if these models are ready for off-the-shelf application to automatically generate abstractive summaries for case judgements. To explore this question, we apply several state-of-the-art domain-specific abstractive summarization models and general-domain LLMs on Indian court case judgements, and check the quality of the generated summaries. In addition to standard metrics for summary quality, we check for inconsistencies and hallucinations in the summaries. We see that abstractive summarization models generally achieve slightly higher scores than extractive models in terms of standard summary evaluation metrics such as ROUGE and BLEU. However, we often find inconsistent or hallucinated information in the generated abstractive summaries. Overall, our investigation indicates that the pre-trained abstractive summarization models and LLMs are not yet ready for fully automatic deployment for case judgement summarization; rather a human-in-the-loop approach including manual checks for inconsistencies is more suitable at present.

Aniket Deroy, Kripabandhu Ghosh, Saptarshi Ghosh

Automatic summarization of legal case judgements, which are known to be long and complex, has traditionally been tried via extractive summarization models. In recent years, generative models including abstractive summarization models and Large language models (LLMs) have gained huge popularity. In this paper, we explore the applicability of such models for legal case judgement summarization. We applied various domain specific abstractive summarization models and general domain LLMs as well as extractive summarization models over two sets of legal case judgements from the United Kingdom (UK) Supreme Court and the Indian (IN) Supreme Court and evaluated the quality of the generated summaries. We also perform experiments on a third dataset of legal documents of a different type, Government reports from the United States (US). Results show that abstractive summarization models and LLMs generally perform better than the extractive methods as per traditional metrics for evaluating summary quality. However, detailed investigation shows the presence of inconsistencies and hallucinations in the outputs of the generative models, and we explore ways to reduce the hallucinations and inconsistencies in the summaries. Overall, the investigation suggests that further improvements are needed to enhance the reliability of abstractive models and LLMs for legal case judgement summarization. At present, a human-in-the-loop technique is more suitable for performing manual checks to identify inconsistencies in the generated summaries.

Shubham Kumar Nigam, Aniket Deroy, Noel Shallum et al.

This paper describes our submission to the SemEval-2023 for Task 6 on LegalEval: Understanding Legal Texts. Our submission concentrated on three subtasks: Legal Named Entity Recognition (L-NER) for Task-B, Legal Judgment Prediction (LJP) for Task-C1, and Court Judgment Prediction with Explanation (CJPE) for Task-C2. We conducted various experiments on these subtasks and presented the results in detail, including data statistics and methodology. It is worth noting that legal tasks, such as those tackled in this research, have been gaining importance due to the increasing need to automate legal analysis and support. Our team obtained competitive rankings of 15$^{th}$, 11$^{th}$, and 1$^{st}$ in Task-B, Task-C1, and Task-C2, respectively, as reported on the leaderboard.

Anurag Roy, Vinay Kumar Verma, Sravan Voonna et al.

Continual Learning (CL) involves training a machine learning model in a sequential manner to learn new information while retaining previously learned tasks without the presence of previous training data. Although there has been significant interest in CL, most recent CL approaches in computer vision have focused on convolutional architectures only. However, with the recent success of vision transformers, there is a need to explore their potential for CL. Although there have been some recent CL approaches for vision transformers, they either store training instances of previous tasks or require a task identifier during test time, which can be limiting. This paper proposes a new exemplar-free approach for class/task incremental learning called ConTraCon, which does not require task-id to be explicitly present during inference and avoids the need for storing previous training instances. The proposed approach leverages the transformer architecture and involves re-weighting the key, query, and value weights of the multi-head self-attention layers of a transformer trained on a similar task. The re-weighting is done using convolution, which enables the approach to maintain low parameter requirements per task. Additionally, an image augmentation-based entropic task identification approach is used to predict tasks without requiring task-ids during inference. Experiments on four benchmark datasets demonstrate that the proposed approach outperforms several competitive approaches while requiring fewer parameters.

Amartya Roy, Elamparithy M, Kripabandhu Ghosh et al.

In context learning (ICL) underpins recent advances in large language models (LLMs), although its role and performance in causal reasoning remains unclear. Causal reasoning demands multihop composition and strict conjunctive control, and reliance on spurious lexical relations of the input could provide misleading results. We hypothesize that, due to their ability to project the input into a latent space, encoder and encoder decoder architectures are better suited for said multihop conjunctive reasoning versus decoder only models. To do this, we compare fine-tuned versions of all the aforementioned architectures with zero and few shot ICL in both natural language and non natural language scenarios. We find that ICL alone is insufficient for reliable causal reasoning, often overfocusing on irrelevant input features. In particular, decoder only models are noticeably brittle to distributional shifts, while finetuned encoder and encoder decoder models can generalize more robustly across our tests, including the non natural language split. Both architectures are only matched or surpassed by decoder only architectures at large scales. We conclude by noting that for cost effective, short horizon robust causal reasoning, encoder or encoder decoder architectures with targeted finetuning are preferable.

Yash Jakhmola, Madhurima Panja, Nitish Kumar Mishra et al.

Spatiotemporal forecasting of traffic flow data represents a typical problem in the field of machine learning, impacting urban traffic management systems. In general, spatiotemporal forecasting problems involve complex interactions, nonlinearities, and long-range dependencies due to the interwoven nature of the temporal and spatial dimensions. Due to this, traditional statistical and machine learning methods cannot adequately handle the temporal and spatial dependencies in these complex traffic flow datasets. A prevalent approach in the field combines graph convolutional networks and multi-head attention mechanisms for spatiotemporal processing. This paper proposes a wavelet-based temporal attention model, namely a wavelet-based dynamic spatiotemporal aware graph neural network (W-DSTAGNN), for tackling the traffic forecasting problem. Wavelet decomposition can help by decomposing the signal into components that can be analyzed independently, reducing the impact of non-stationarity and handling long-range dependencies of traffic flow datasets. Benchmark experiments using three popularly used statistical metrics confirm that our proposal efficiently captures spatiotemporal correlations and outperforms ten state-of-the-art models (including both temporal and spatiotemporal benchmarks) on three publicly available traffic datasets. Our proposed ensemble method can better handle dynamic temporal and spatial dependencies and make reliable long-term forecasts. In addition to point forecasts, our proposed model can generate interval forecasts that significantly enhance probabilistic forecasting for traffic datasets.

Shubham Kumar Nigam, Tanuj Tyagi, Siddharth Shukla et al.

This paper presents an early exploration of reinforcement learning methodologies for legal AI in the Indian context. We introduce Reinforcement Learning-based Legal Reasoning (ReGal), a framework that integrates Multi-Task Instruction Tuning with Reinforcement Learning from AI Feedback (RLAIF) using Proximal Policy Optimization (PPO). Our approach is evaluated across two critical legal tasks: (i) Court Judgment Prediction and Explanation (CJPE), and (ii) Legal Document Summarization. Although the framework underperforms on standard evaluation metrics compared to supervised and proprietary models, it provides valuable insights into the challenges of applying RL to legal texts. These challenges include reward model alignment, legal language complexity, and domain-specific adaptation. Through empirical and qualitative analysis, we demonstrate how RL can be repurposed for high-stakes, long-document tasks in law. Our findings establish a foundation for future work on optimizing legal reasoning pipelines using reinforcement learning, with broader implications for building interpretable and adaptive legal AI systems.

Amartya Roy, Danush Khanna, Devanshu Mahapatra et al.

This paper tackles the challenge of building robust and generalizable bias mitigation models for language. Recognizing the limitations of existing datasets, we introduce ANUBIS, a novel dataset with 1507 carefully curated sentence pairs encompassing nine social bias categories. We evaluate state-of-the-art models like T5, utilizing Supervised Fine-Tuning (SFT), Reinforcement Learning (PPO, DPO), and In-Context Learning (ICL) for effective bias mitigation. Our analysis focuses on multi-class social bias reduction, cross-dataset generalizability, and environmental impact of the trained models. ANUBIS and our findings offer valuable resources for building more equitable AI systems and contribute to the development of responsible and unbiased technologies with broad societal impact.

Danush Khanna, Pratinav Seth, Sidhaarth Sredharan Murali et al.

Mental manipulation is a subtle yet pervasive form of abuse in interpersonal communication, making its detection critical for safeguarding potential victims. However, due to manipulation's nuanced and context-specific nature, identifying manipulative language in complex, multi-turn, and multi-person conversations remains a significant challenge for large language models (LLMs). To address this gap, we introduce the MultiManip dataset, comprising 220 multi-turn, multi-person dialogues balanced between manipulative and non-manipulative interactions, all drawn from reality shows that mimic real-world scenarios. For manipulative interactions, it includes 11 distinct manipulations depicting real-life scenarios. We conduct extensive evaluations of state-of-the-art LLMs, such as GPT-4o and Llama-3.1-8B, employing various prompting strategies. Despite their capabilities, these models often struggle to detect manipulation effectively. To overcome this limitation, we propose SELF-PERCEPT, a novel, two-stage prompting framework inspired by Self-Perception Theory, demonstrating strong performance in detecting multi-person, multi-turn mental manipulation. Our code and data are publicly available at https://github.com/danushkhanna/self-percept .

Shubham Kumar Nigam, Balaramamahanthi Deepak Patnaik, Ajay Varghese Thomas et al.

Automating legal document drafting can significantly enhance efficiency, reduce manual effort, and streamline legal workflows. While prior research has explored tasks such as judgment prediction and case summarization, the structured generation of private legal documents in the Indian legal domain remains largely unaddressed. To bridge this gap, we introduce VidhikDastaavej, a novel, anonymized dataset of private legal documents, and develop NyayaShilp, a fine-tuned legal document generation model specifically adapted to Indian legal texts. We propose a Model-Agnostic Wrapper (MAW), a two-step framework that first generates structured section titles and then iteratively produces content while leveraging retrieval-based mechanisms to ensure coherence and factual accuracy. We benchmark multiple open-source LLMs, including instruction-tuned and domain-adapted versions, alongside proprietary models for comparison. Our findings indicate that while direct fine-tuning on small datasets does not always yield improvements, our structured wrapper significantly enhances coherence, factual adherence, and overall document quality while mitigating hallucinations. To ensure real-world applicability, we developed a Human-in-the-Loop (HITL) Document Generation System, an interactive user interface that enables users to specify document types, refine section details, and generate structured legal drafts. This tool allows legal professionals and researchers to generate, validate, and refine AI-generated legal documents efficiently. Extensive evaluations, including expert assessments, confirm that our framework achieves high reliability in structured legal drafting. This research establishes a scalable and adaptable foundation for AI-assisted legal drafting in India, offering an effective approach to structured legal document generation.

Anurag Roy, Shalmoli Ghosh, Kripabandhu Ghosh et al.

A large fraction of textual data available today contains various types of 'noise', such as OCR noise in digitized documents, noise due to informal writing style of users on microblogging sites, and so on. To enable tasks such as search/retrieval and classification over all the available data, we need robust algorithms for text normalization, i.e., for cleaning different kinds of noise in the text. There have been several efforts towards cleaning or normalizing noisy text; however, many of the existing text normalization methods are supervised and require language-dependent resources or large amounts of training data that is difficult to obtain. We propose an unsupervised algorithm for text normalization that does not need any training data / human intervention. The proposed algorithm is applicable to text over different languages, and can handle both machine-generated and human-generated noise. Experiments over several standard datasets show that text normalization through the proposed algorithm enables better retrieval and stance detection, as compared to that using several baseline text normalization methods. Implementation of our algorithm can be found at https://github.com/ranarag/UnsupClean.

Soham Wasmatkar, Subinay Adhikary, Rakshit Rohan et al.

In 2011, Judea Pearl received the Turing Award, considered the Nobel Prize in Computing, for fundamental contributions to artificial intelligence through the development of a calculus for probabilistic and causal reasoning. It includes pioneering the development of causal discovery algorithms. These computer algorithms can analyze large multivariate datasets and automatically discover the causal relationships among the constituent variables. They have been widely used in many critical fields such as medicine and economics to support decisions. However, to our knowledge, they have not been leveraged in law. This paper attempts to alleviate this gap by investigating whether causal discovery algorithms can be leveraged for automated generation of legal arguments. To that end, a novel legal dataset is prepared by identifying 17 legal concepts, such as physical assault and property dispute. A curated collection of 150 homicide cases are annotated with these concepts, e.g., a case is annotated with physical assault only if a physical assault had been reported in that case. Subsequently, a selected set of widely-used causal discovery algorithms is applied to the annotated dataset to discover the causal relationships between the legal concepts. Additionally, the degrees of belief associated with the discovered relationships are quantified in mathematical probabilities. It is shown that some of the causal relationships help generate viable legal arguments, e.g., if one could establish that a physical assault has not taken place during a homicide, it should be a sufficient condition (with probability 1) to establish that the homicide has not been committed due to a property-related dispute. Thus, this paper shows that causal discovery algorithms can be helpful in generating legal arguments, opening up avenues for promising future endeavors.

Garima Chhikara, Anurag Sharma, Kripabandhu Ghosh et al.

Employing Large Language Models (LLM) in various downstream applications such as classification is crucial, especially for smaller companies lacking the expertise and resources required for fine-tuning a model. Fairness in LLMs helps ensure inclusivity, equal representation based on factors such as race, gender and promotes responsible AI deployment. As the use of LLMs has become increasingly prevalent, it is essential to assess whether LLMs can generate fair outcomes when subjected to considerations of fairness. In this study, we introduce a framework outlining fairness regulations aligned with various fairness definitions, with each definition being modulated by varying degrees of abstraction. We explore the configuration for in-context learning and the procedure for selecting in-context demonstrations using RAG, while incorporating fairness rules into the process. Experiments conducted with different LLMs indicate that GPT-4 delivers superior results in terms of both accuracy and fairness compared to other models. This work is one of the early attempts to achieve fairness in prediction tasks by utilizing LLMs through in-context learning.

Aniket Deroy, Naksatra Kumar Bailung, Kripabandhu Ghosh et al.

Despite the availability of vast amounts of data, legal data is often unstructured, making it difficult even for law practitioners to ingest and comprehend the same. It is important to organise the legal information in a way that is useful for practitioners and downstream automation tasks. The word ontology was used by Greek philosophers to discuss concepts of existence, being, becoming and reality. Today, scientists use this term to describe the relation between concepts, data, and entities. A great example for a working ontology was developed by Dhani and Bhatt. This ontology deals with Indian court cases on intellectual property rights (IPR) The future of legal ontologies is likely to be handled by computer experts and legal experts alike.

Shubham Kumar Nigam, Tanmay Dubey, Govind Sharma et al.

In this paper, we address the task of semantic segmentation of legal documents through rhetorical role classification, with a focus on Indian legal judgments. We introduce LegalSeg, the largest annotated dataset for this task, comprising over 7,000 documents and 1.4 million sentences, labeled with 7 rhetorical roles. To benchmark performance, we evaluate multiple state-of-the-art models, including Hierarchical BiLSTM-CRF, TransformerOverInLegalBERT (ToInLegalBERT), Graph Neural Networks (GNNs), and Role-Aware Transformers, alongside an exploratory RhetoricLLaMA, an instruction-tuned large language model. Our results demonstrate that models incorporating broader context, structural relationships, and sequential sentence information outperform those relying solely on sentence-level features. Additionally, we conducted experiments using surrounding context and predicted or actual labels of neighboring sentences to assess their impact on classification accuracy. Despite these advancements, challenges persist in distinguishing between closely related roles and addressing class imbalance. Our work underscores the potential of advanced techniques for improving legal document understanding and sets a strong foundation for future research in legal NLP.

Ayushman Gupta, Akhil Bhogal, Kripabandhu Ghosh

Code-mixing, the practice of alternating between two or more languages in an utterance, is a common phenomenon in multilingual communities. Due to the colloquial nature of code-mixing, there is no singular correct way to translate an English sentence into a code-mixed sentence. For this reason, standard n-gram-based MT evaluation metrics such as the BLEU score are not appropriate for code-mixed evaluation. To demonstrate this, we propose a novel method for code-mixed text generation: Controlled Generation, which parameterizes the code-mixing degree (CMD) and enables the generation of multiple semantically equivalent code-mixed sentences from a given English sentence. We introduce a robust new evaluation metric: GAME: A Gold-Standard Agnostic Measure for Evaluation of Code-Mixed Sentences. GAME is both language-agnostic and gold-standard-agnostic, i.e. unlike other metrics, GAME does not require gold-standard code-mixed sentences for evaluation, thus eliminating the need for human annotators in the code-mixed evaluation process. When used to evaluate semantically equivalent code-mixed sentences, we find that GAME scores have a lower standard deviation than BLEU scores. Further, we create and release a dataset containing gold-standard code-mixed sentences across 4 language pairs: English-{Hindi, Bengali, French, Spanish} to encourage more computational research on code-mixing.

Purbid Bambroo, Subinay Adhikary, Paheli Bhattacharya et al.

Identification of rhetorical roles like facts, arguments, and final judgments is central to understanding a legal case document and can lend power to other downstream tasks like legal case summarization and judgment prediction. However, there are several challenges to this task. Legal documents are often unstructured and contain a specialized vocabulary, making it hard for conventional transformer models to understand them. Additionally, these documents run into several pages, which makes it difficult for neural models to capture the entire context at once. Lastly, there is a dearth of annotated legal documents to train deep learning models. Previous state-of-the-art approaches for this task have focused on using neural models like BiLSTM-CRF or have explored different embedding techniques to achieve decent results. While such techniques have shown that better embedding can result in improved model performance, not many models have focused on utilizing attention for learning better embeddings in sentences of a document. Additionally, it has been recently shown that advanced techniques like multi-task learning can help the models learn better representations, thereby improving performance. In this paper, we combine these two aspects by proposing a novel family of multi-task learning-based models for rhetorical role labeling, named MARRO, that uses transformer-inspired multi-headed attention. Using label shift as an auxiliary task, we show that models from the MARRO family achieve state-of-the-art results on two labeled datasets for rhetorical role labeling, from the Indian and UK Supreme Courts.

Shubham Kumar Nigam, Balaramamahanthi Deepak Patnaik, Shivam Mishra et al.

The integration of artificial intelligence (AI) in legal judgment prediction (LJP) has the potential to transform the legal landscape, particularly in jurisdictions like India, where a significant backlog of cases burdens the legal system. This paper introduces NyayaAnumana, the largest and most diverse corpus of Indian legal cases compiled for LJP, encompassing a total of 7,02,945 preprocessed cases. NyayaAnumana, which combines the words "Nyay" (judgment) and "Anuman" (prediction or inference) respectively for most major Indian languages, includes a wide range of cases from the Supreme Court, High Courts, Tribunal Courts, District Courts, and Daily Orders and, thus, provides unparalleled diversity and coverage. Our dataset surpasses existing datasets like PredEx and ILDC, offering a comprehensive foundation for advanced AI research in the legal domain. In addition to the dataset, we present INLegalLlama, a domain-specific generative large language model (LLM) tailored to the intricacies of the Indian legal system. It is developed through a two-phase training approach over a base LLaMa model. First, Indian legal documents are injected using continual pretraining. Second, task-specific supervised finetuning is done. This method allows the model to achieve a deeper understanding of legal contexts. Our experiments demonstrate that incorporating diverse court data significantly boosts model accuracy, achieving approximately 90% F1-score in prediction tasks. INLegalLlama not only improves prediction accuracy but also offers comprehensible explanations, addressing the need for explainability in AI-assisted legal decisions.

Ayushman Gupta, Akhil Bhogal, Kripabandhu Ghosh

Multilingual Large Language Models (LLMs) have demonstrated exceptional performance in Machine Translation (MT) tasks. However, their MT abilities in the context of code-switching (the practice of mixing two or more languages in an utterance) remain under-explored. In this paper, we introduce Rule-Based Prompting, a novel prompting technique to generate code-mixed sentences. We measure and compare the code-mixed MT abilities of 3 popular multilingual LLMs: GPT-3.5-turbo, GPT-4, and Gemini Pro across five language pairs: English-{Hindi, Bengali, Gujarati, French, Spanish} using $k$-shot prompting ($k\in\{0, 1, 10, 20\}$) and Rule-Based Prompting. Our findings suggest that though $k$-shot prompting often leads to the best results, Rule-Based prompting shows promise in generating unique code-mixed sentences that vary in their style of code-mixing. We also use $k$-shot prompting to gauge the code-mixed to English translation abilities of multilingual LLMs. For this purpose, we create a gold-standard code-mixed dataset spanning five language pairs: English-{Hindi, Bengali, Gujarati, French, Spanish}. As a real-world application of our work, we create a code-mixed chatbot.

Shubham Kumar Nigam, Balaramamahanthi Deepak Patnaik, Shivam Mishra et al.

Legal Judgment Prediction (LJP) has emerged as a key area in AI for law, aiming to automate judicial outcome forecasting and enhance interpretability in legal reasoning. While previous approaches in the Indian context have relied on internal case content such as facts, issues, and reasoning, they often overlook a core element of common law systems, which is reliance on statutory provisions and judicial precedents. In this work, we propose NyayaRAG, a Retrieval-Augmented Generation (RAG) framework that simulates realistic courtroom scenarios by providing models with factual case descriptions, relevant legal statutes, and semantically retrieved prior cases. NyayaRAG evaluates the effectiveness of these combined inputs in predicting court decisions and generating legal explanations using a domain-specific pipeline tailored to the Indian legal system. We assess performance across various input configurations using both standard lexical and semantic metrics as well as LLM-based evaluators such as G-Eval. Our results show that augmenting factual inputs with structured legal knowledge significantly improves both predictive accuracy and explanation quality.

Shubham Kumar Nigam, Balaramamahanthi Deepak Patnaik, Shivam Mishra et al.

In the landscape of Fact-based Judgment Prediction and Explanation (FJPE), reliance on factual data is essential for developing robust and realistic AI-driven decision-making tools. This paper introduces TathyaNyaya, the largest annotated dataset for FJPE tailored to the Indian legal context, encompassing judgments from the Supreme Court of India and various High Courts. Derived from the Hindi terms "Tathya" (fact) and "Nyaya" (justice), the TathyaNyaya dataset is uniquely designed to focus on factual statements rather than complete legal texts, reflecting real-world judicial processes where factual data drives outcomes. Complementing this dataset, we present FactLegalLlama, an instruction-tuned variant of the LLaMa-3-8B Large Language Model (LLM), optimized for generating high-quality explanations in FJPE tasks. Finetuned on the factual data in TathyaNyaya, FactLegalLlama integrates predictive accuracy with coherent, contextually relevant explanations, addressing the critical need for transparency and interpretability in AI-assisted legal systems. Our methodology combines transformers for binary judgment prediction with FactLegalLlama for explanation generation, creating a robust framework for advancing FJPE in the Indian legal domain. TathyaNyaya not only surpasses existing datasets in scale and diversity but also establishes a benchmark for building explainable AI systems in legal analysis. The findings underscore the importance of factual precision and domain-specific tuning in enhancing predictive performance and interpretability, positioning TathyaNyaya and FactLegalLlama as foundational resources for AI-assisted legal decision-making.

Abhilekh Borah, Shubhra Ghosh, Kedar Joshi et al.

Tasks such as solving arithmetic equations, evaluating truth tables, and completing syllogisms are handled well by large language models (LLMs) in their standard form, but they often fail when the same problems are posed in logically equivalent yet obfuscated formats. To study this vulnerability, we introduce Logifus, a structure-preserving logical obfuscation framework, and, utilizing this, we present LogiQAte, a first-of-its-kind diagnostic benchmark with 1,108 questions across four reasoning tasks: (i) Obfus FOL (first-order logic entailment under equivalence-preserving rewrites), (ii) Obfus Blood Relation (family-graph entailment under indirect relational chains), (iii) Obfus Number Series (pattern induction under symbolic substitutions), and (iv) Obfus Direction Sense (navigation reasoning under altered directions and reference frames). Across all the tasks, evaluating six state-of-the-art models, we find that obfuscation severely degrades zero-shot performance, with performance dropping on average by 47% for GPT-4o, 27% for GPT-5, and 22% for reasoning model, o4-mini. Our findings reveal that current LLMs parse questions without deep understanding, highlighting the urgency of building models that genuinely comprehend and preserve meaning beyond surface form.

Srivarshinee Sridhar, Raghav Kaushik Ravi, Kripabandhu Ghosh

Large Language Models (LLMs) are increasingly used in clinical settings, where sensitivity to linguistic uncertainty can influence diagnostic interpretation and decision-making. Yet little is known about where such epistemic cues are internally represented within these models. Distinct from uncertainty quantification, which measures output confidence, this work examines input-side representational sensitivity to linguistic uncertainty in medical text. We curate a contrastive dataset of clinical statements varying in epistemic modality (e.g., 'is consistent with' vs. 'may be consistent with') and propose Model Sensitivity to Uncertainty (MSU), a layerwise probing metric that quantifies activation-level shifts induced by uncertainty cues. Our results show that LLMs exhibit structured, depth-dependent sensitivity to clinical uncertainty, suggesting that epistemic information is progressively encoded in deeper layers. These findings reveal how linguistic uncertainty is internally represented in LLMs, offering insight into their interpretability and epistemic reliability.

Amartya Roy, Devharish N, Shreya Ganguly et al.

Modern causal discovery methods face critical limitations in scalability, computational efficiency, and adaptability to mixed data types, as evidenced by benchmarks on node scalability (30, $\le 50$, $\ge 70$ nodes), computational energy demands, and continuous/non-continuous data handling. While traditional algorithms like PC, GES, and ICA-LiNGAM struggle with these challenges, exhibiting prohibitive energy costs for higher-order nodes and poor scalability beyond 70 nodes, we propose \textbf{GUIDE}, a framework that integrates Large Language Model (LLM)-generated adjacency matrices with observational data through a dual-encoder architecture. GUIDE uniquely optimizes computational efficiency, reducing runtime on average by $\approx 42%$ compared to RL-BIC and KCRL methods, while achieving an average $\approx 117%$ improvement in accuracy over both NOTEARS and GraN-DAG individually. During training, GUIDE's reinforcement learning agent dynamically balances reward maximization (accuracy) and penalty avoidance (DAG constraints), enabling robust performance across mixed data types and scalability to $\ge 70$ nodes -- a setting where baseline methods fail.

Garima Chhikara, Kripabandhu Ghosh, Abhijnan Chakraborty

Large Language Models (LLMs) are widely used for downstream tasks such as tabular classification, where ensuring fairness in their outputs is critical for inclusivity, equal representation, and responsible AI deployment. This study introduces a novel approach to enhancing LLM performance and fairness through the concept of a dynamic validation set, which evolves alongside the test set, replacing the traditional static validation approach. We also propose an iterative algorithm, SMITE, to select optimal in-context examples, with each example set validated against its corresponding dynamic validation set. The in-context set with the lowest total error is used as the final demonstration set. Our experiments across four different LLMs show that our proposed techniques significantly improve both predictive accuracy and fairness compared to baseline methods. To our knowledge, this is the first study to apply dynamic validation in the context of in-context learning for LLMs.

Shubhra Ghosh, Abhilekh Borah, Aditya Kumar Guru et al.

The rapid proliferation of Large Language Models (LLMs) has significantly contributed to the development of equitable AI systems capable of factual question-answering (QA). However, no known study tests the LLMs' robustness when presented with obfuscated versions of questions. To systematically evaluate these limitations, we propose a novel technique, ObfusQAte and, leveraging the same, introduce ObfusQA, a comprehensive, first of its kind, framework with multi-tiered obfuscation levels designed to examine LLM capabilities across three distinct dimensions: (i) Named-Entity Indirection, (ii) Distractor Indirection, and (iii) Contextual Overload. By capturing these fine-grained distinctions in language, ObfusQA provides a comprehensive benchmark for evaluating LLM robustness and adaptability. Our study observes that LLMs exhibit a tendency to fail or generate hallucinated responses when confronted with these increasingly nuanced variations. To foster research in this direction, we make ObfusQAte publicly available.

Danush Khanna, Aditya Kumar Guru, Srivarshinee Sridhar et al.

Inference accounts for the majority of latency and energy consumption in large language model (LLM) deployments, often exceeding 90% of total cost. While training-time efficiency has seen extensive progress, runtime optimization remains a key bottleneck, particularly under autoregressive decoding. Existing approaches -- such as pruning, quantization, early exits, and speculative decoding -- often require retraining, architectural changes, or disrupt decoding compatibility. We introduce QuickSilver, a modular, token-level framework that enables semantic adaptivity at inference time without altering model weights or structure. QuickSilver integrates four synergistic mechanisms: (i) Dynamic Token Halting, which halts computation for tokens with converged representations; (ii) KV Cache Skipping, which selectively suppresses memory writes to reduce attention overhead; and (iii) Contextual Token Fusion, which collapses redundant tokens into shared paths to shrink sequence length. Unlike speculative decoding or MoE routing, QuickSilver operates entirely on frozen, dense models and requires no auxiliary networks. Applied to GPT-2 and Llama-2 across WikiText-103 and C4, QuickSilver achieves up to 39.6% FLOP reduction with negligible perplexity degradation (<=0.2).

Arion Das, Asutosh Mishra, Amitesh Patel et al.

Large Language Models (LLMs) have significantly impacted nearly every domain of human knowledge. However, the explainability of these models esp. to laypersons, which are crucial for instilling trust, have been examined through various skeptical lenses. In this paper, we introduce a novel notion of LLM explainability to laypersons, termed $\textit{ReQuesting}$, across three high-priority application domains -- law, health and finance, using multiple state-of-the-art LLMs. The proposed notion exhibits faithful generation of explainable layman-understandable algorithms on multiple tasks through high degree of reproducibility. Furthermore, we observe a notable alignment of the explainable algorithms with intrinsic reasoning of the LLMs.

Garima Chhikara, Anurag Sharma, V. Gurucharan et al.

Citizen reporting platforms like Safe City in India help the public and authorities stay informed about sexual harassment incidents. However, the high volume of data shared on these platforms makes reviewing each individual case challenging. Therefore, a summarization algorithm capable of processing and understanding various Indian code-mixed languages is essential. In recent years, Large Language Models (LLMs) have shown exceptional performance in NLP tasks, including summarization. LLMs inherently produce abstractive summaries by paraphrasing the original text, while the generation of extractive summaries - selecting specific subsets from the original text - through LLMs remains largely unexplored. Moreover, LLMs have a limited context window size, restricting the amount of data that can be processed at once. We tackle these challenge by introducing LaMSUM, a novel multi-level framework designed to generate extractive summaries for large collections of Safe City posts using LLMs. LaMSUM integrates summarization with different voting methods to achieve robust summaries. Extensive evaluation using three popular LLMs (Llama, Mistral and GPT-4o) demonstrates that LaMSUM outperforms state-of-the-art extractive summarization methods for Safe City posts. Overall, this work represents one of the first attempts to achieve extractive summarization through LLMs, and is likely to support stakeholders by offering a comprehensive overview and enabling them to develop effective policies to minimize incidents of unwarranted harassment.

Shubham Kumar Nigam, Anurag Sharma, Danush Khanna et al.

In the era of Large Language Models (LLMs), predicting judicial outcomes poses significant challenges due to the complexity of legal proceedings and the scarcity of expert-annotated datasets. Addressing this, we introduce \textbf{Pred}iction with \textbf{Ex}planation (\texttt{PredEx}), the largest expert-annotated dataset for legal judgment prediction and explanation in the Indian context, featuring over 15,000 annotations. This groundbreaking corpus significantly enhances the training and evaluation of AI models in legal analysis, with innovations including the application of instruction tuning to LLMs. This method has markedly improved the predictive accuracy and explanatory depth of these models for legal judgments. We employed various transformer-based models, tailored for both general and Indian legal contexts. Through rigorous lexical, semantic, and expert assessments, our models effectively leverage \texttt{PredEx} to provide precise predictions and meaningful explanations, establishing it as a valuable benchmark for both the legal profession and the NLP community.

Paheli Bhattacharya, Soham Poddar, Koustav Rudra et al.

Automatic summarization of legal case documents is an important and practical challenge. Apart from many domain-independent text summarization algorithms that can be used for this purpose, several algorithms have been developed specifically for summarizing legal case documents. However, most of the existing algorithms do not systematically incorporate domain knowledge that specifies what information should ideally be present in a legal case document summary. To address this gap, we propose an unsupervised summarization algorithm DELSumm which is designed to systematically incorporate guidelines from legal experts into an optimization setup. We conduct detailed experiments over case documents from the Indian Supreme Court. The experiments show that our proposed unsupervised method outperforms several strong baselines in terms of ROUGE scores, including both general summarization algorithms and legal-specific ones. In fact, though our proposed algorithm is unsupervised, it outperforms several supervised summarization models that are trained over thousands of document-summary pairs.

Anurag Shandilya, Abhisek Dash, Abhijnan Chakraborty et al.

With the surge in user-generated textual information, there has been a recent increase in the use of summarization algorithms for providing an overview of the extensive content. Traditional metrics for evaluation of these algorithms (e.g. ROUGE scores) rely on matching algorithmic summaries to human-generated ones. However, it has been shown that when the textual contents are heterogeneous, e.g., when they come from different socially salient groups, most existing summarization algorithms represent the social groups very differently compared to their distribution in the original data. To mitigate such adverse impacts, some fairness-preserving summarization algorithms have also been proposed. All of these studies have considered normative notions of fairness from the perspective of writers of the contents, neglecting the readers' perceptions of the underlying fairness notions. To bridge this gap, in this work, we study the interplay between the fairness notions and how readers perceive them in textual summaries. Through our experiments, we show that reader's perception of fairness is often context-sensitive. Moreover, standard ROUGE evaluation metrics are unable to quantify the perceived (un)fairness of the summaries. To this end, we propose a human-in-the-loop metric and an automated graph-based methodology to quantify the perceived bias in textual summaries. We demonstrate their utility by quantifying the (un)fairness of several summaries of heterogeneous socio-political microblog datasets.

Anurag Roy, Vinay Kumar Verma, Kripabandhu Ghosh et al.

Most existing algorithms for cross-modal Information Retrieval are based on a supervised train-test setup, where a model learns to align the mode of the query (e.g., text) to the mode of the documents (e.g., images) from a given training set. Such a setup assumes that the training set contains an exhaustive representation of all possible classes of queries. In reality, a retrieval model may need to be deployed on previously unseen classes, which implies a zero-shot IR setup. In this paper, we propose a novel GAN-based model for zero-shot text to image retrieval. When given a textual description as the query, our model can retrieve relevant images in a zero-shot setup. The proposed model is trained using an Expectation-Maximization framework. Experiments on multiple benchmark datasets show that our proposed model comfortably outperforms several state-of-the-art zero-shot text to image retrieval models, as well as zero-shot classification and hashing models suitably used for retrieval.

Ritam Dutt, Moumita Basu, Kripabandhu Ghosh et al.

During a disaster event, two types of information that are especially useful for coordinating relief operations are needs and availabilities of resources (e.g., food, water, medicines) in the affected region. Information posted on microblogging sites is increasingly being used for assisting post-disaster relief operations. In this context, two practical challenges are (i)~to identify tweets that inform about resource needs and availabilities (termed as need-tweets and availability-tweets respectively), and (ii)~to automatically match needs with appropriate availabilities. While several works have addressed the first problem, there has been little work on automatically matching needs with availabilities. The few prior works that attempted matching only considered the resources, and no attempt has been made to understand other aspects of needs/availabilities that are essential for matching in practice. In this work, we develop a methodology for understanding five important aspects of need-tweets and availability-tweets, including what resource and what quantity is needed/available, the geographical location of the need/availability, and who needs / is providing the resource. Understanding these aspects helps us to address the need-availability matching problem considering not only the resources, but also other factors such as the geographical proximity between the need and the availability. To our knowledge, this study is the first attempt to develop methods for understanding the semantics of need-tweets and availability-tweets. We also develop a novel methodology for matching need-tweets with availability-tweets, considering both resource similarity and geographical proximity. Experiments on two datasets corresponding to two disaster events, demonstrate that our proposed methods perform substantially better matching than those in prior works.

Paheli Bhattacharya, Kripabandhu Ghosh, Arindam Pal et al.

Computing similarity between two legal case documents is an important and challenging task in Legal IR, for which text-based and network-based measures have been proposed in literature. All prior network-based similarity methods considered a precedent citation network among case documents only (PCNet). However, this approach misses an important source of legal knowledge -- the hierarchy of legal statutes that are applicable in a given legal jurisdiction (e.g., country). We propose to augment the PCNet with the hierarchy of legal statutes, to form a heterogeneous network Hier-SPCNet, having citation links between case documents and statutes, as well as citation and hierarchy links among the statutes. Experiments over a set of Indian Supreme Court case documents show that our proposed heterogeneous network enables significantly better document similarity estimation, as compared to existing approaches using PCNet. We also show that the proposed network-based method can complement text-based measures for better estimation of legal document similarity.

Kaustubh Hiware, Ritam Dutt, Sayan Sinha et al.

Although a lot of research has been done on utilising Online Social Media during disasters, there exists no system for a specific task that is critical in a post-disaster scenario -- identifying resource-needs and resource-availabilities in the disaster-affected region, coupled with their subsequent matching. To this end, we present NARMADA, a semi-automated platform which leverages the crowd-sourced information from social media posts for assisting post-disaster relief coordination efforts. The system employs Natural Language Processing and Information Retrieval techniques for identifying resource-needs and resource-availabilities from microblogs, extracting resources from the posts, and also matching the needs to suitable availabilities. The system is thus capable of facilitating the judicious management of resources during post-disaster relief operations.

Paheli Bhattacharya, Kripabandhu Ghosh, Arindam Pal et al.

Computing similarity between two legal documents is an important and challenging task in the domain of Legal Information Retrieval. Finding similar legal documents has many applications in downstream tasks, including prior-case retrieval, recommendation of legal articles, and so on. Prior works have proposed two broad ways of measuring similarity between legal documents - analyzing the precedent citation network, and measuring similarity based on textual content similarity measures. But there has not been a comprehensive comparison of these existing methods on a common platform. In this paper, we perform the first systematic analysis of the existing methods. In addition, we explore two promising new similarity computation methods - one text-based and the other based on network embeddings, which have not been considered till now.

Paheli Bhattacharya, Shounak Paul, Kripabandhu Ghosh et al.

Automatically understanding the rhetorical roles of sentences in a legal case judgement is an important problem to solve, since it can help in several downstream tasks like summarization of legal judgments, legal search, and so on. The task is challenging since legal case documents are usually not well-structured, and these rhetorical roles may be subjective (as evident from variation of opinions between legal experts). In this paper, we address this task for judgments from the Supreme Court of India. We label sentences in 50 documents using multiple human annotators, and perform an extensive analysis of the human-assigned labels. We also attempt automatic identification of the rhetorical roles of sentences. While prior approaches towards this task used Conditional Random Fields over manually handcrafted features, we explore the use of deep neural models which do not require hand-crafting of features. Experiments show that neural models perform much better in this task than baseline methods which use handcrafted features.

Abhisek Dash, Anurag Shandilya, Arindam Biswas et al.

As the amount of user-generated textual content grows rapidly, text summarization algorithms are increasingly being used to provide users a quick overview of the information content. Traditionally, summarization algorithms have been evaluated only based on how well they match human-written summaries (e.g. as measured by ROUGE scores). In this work, we propose to evaluate summarization algorithms from a completely new perspective that is important when the user-generated data to be summarized comes from different socially salient user groups, e.g. men or women, Caucasians or African-Americans, or different political groups (Republicans or Democrats). In such cases, we check whether the generated summaries fairly represent these different social groups. Specifically, considering that an extractive summarization algorithm selects a subset of the textual units (e.g. microblogs) in the original data for inclusion in the summary, we investigate whether this selection is fair or not. Our experiments over real-world microblog datasets show that existing summarization algorithms often represent the socially salient user-groups very differently compared to their distributions in the original data. More importantly, some groups are frequently under-represented in the generated summaries, and hence get far less exposure than what they would have obtained in the original data. To reduce such adverse impacts, we propose novel fairness-preserving summarization algorithms which produce high-quality summaries while ensuring fairness among various groups. To our knowledge, this is the first attempt to produce fair text summarization, and is likely to open up an interesting research direction.

Mitodru Niyogi, Kripabandhu Ghosh, Arnab Bhattacharya

Cross-lingual information retrieval is a challenging task in the absence of aligned parallel corpora. In this paper, we address this problem by considering topically aligned corpora designed for evaluating an IR setup. To emphasize, we neither use any sentence-aligned corpora or document-aligned corpora, nor do we use any language specific resources such as dictionary, thesaurus, or grammar rules. Instead, we use an embedding into a common space and learn word correspondences directly from there. We test our proposed approach for bilingual IR on standard FIRE datasets for Bangla, Hindi and English. The proposed method is superior to the state-of-the-art method not only for IR evaluation measures but also in terms of time requirements. We extend our method successfully to the trilingual setting.

Prannay Khosla, Moumita Basu, Kripabandhu Ghosh et al.

Microblogging sites like Twitter and Weibo have emerged as important sourcesof real-time information on ongoing events, including socio-political events, emergency events, and so on. For instance, during emergency events (such as earthquakes, floods, terror attacks), microblogging sites are very useful for gathering situational information in real-time. During such an event, typically only a small fraction of the microblogs (tweets) posted are relevant to the information need. Hence, it is necessary to design effective methodologies for microblog retrieval, so that the relevant tweets can be automatically extracted from large sets of documents (tweets). In this work, we apply and compare various neural network-based IR models for microblog retrieval for a specific application, as follows. In a disaster situation, one of the primary and practical challenges in coordinating the post-disaster relief operations is to know about what resources are needed and what resources are available in the disaster-affected area. Thus, in this study, we focus on extracting these two specific types of microblogs or tweets namely need tweets and avail tweets, which are tweets which define some needs of the people and the tweets which offer some solutions or aid for the people, respectively.