Trishita Saha

Author directory

2026

Legal documents are often lengthy and contain domain-specific terminology and complex legal reasoning, making them difficult for the general public to understand. This challenge is further amplified in the Indian context, where a large portion of the population lacks proficiency in complex English. Although automatic legal summarization has been studied, most existing approaches generate summaries intended for legal professionals and thus remain difficult for laypeople. To address this gap, we introduce InLegalLaySum, the first benchmark for lay summarization of Indian legal documents. InLegalLaySum comprises 9,570 judgements from the Supreme Court and several High Courts of India paired with expert-written summaries in English. We benchmark a wide range of generative models (ranging from 160M to 7B parameters) and perform comprehensive experiments with multiple training strategies, including supervised fine-tuning (SFT) and direct preference optimization (DPO). We observed that the hybrid SFT+DPO training achieved superior performance across readability metrics, which is further validated by domain experts.

2025

This paper describes the system created for the BEA 2025 Shared Task on Pedagogical Ability Assessment of AI-powered Tutors. The task aims to assess how well AI tutors identify and locate errors made by students, provide guidance and ensure actionability, among other features of their responses in educational dialogues. Transformer-based models, especially DeBERTa and RoBERTa, are improved by multitask learning, threshold tweaking, ordinal regression, and oversampling. The efficiency of pedagogically driven training methods and bespoke transformer models for evaluating AI tutor quality is demonstrated by the high performance of their best systems across all evaluation tracks.