Scholarlink

This paper proposes a predictive modeling framework for classifying bail outcomes in Indian courts using the HLDC (Hindi Legal Document Corpus) dataset and IndicBERT, a transformer-based language model specialized for Indian languages. The study applies various machine learning and natural language processing (NLP) techniques to predict whether bail is “granted” or “denied.” The methodology utilizes TF-IDF for document summarization, alongside fine-tuning IndicBERT to capture legal-specific patterns and terminologies, thereby improving model accuracy. Extensive hyperparameter tuning further optimizes the model’s performance in legal text classification tasks.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predictive Modeling for Bail Applications in Indian Courts Using IndicBERT and HLDC Dataset

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ananya Sivakumar,
  • Ashuwin Palanivel,
  • Karthika Subbaraj

摘要

This paper proposes a predictive modeling framework for classifying bail outcomes in Indian courts using the HLDC (Hindi Legal Document Corpus) dataset and IndicBERT, a transformer-based language model specialized for Indian languages. The study applies various machine learning and natural language processing (NLP) techniques to predict whether bail is “granted” or “denied.” The methodology utilizes TF-IDF for document summarization, alongside fine-tuning IndicBERT to capture legal-specific patterns and terminologies, thereby improving model accuracy. Extensive hyperparameter tuning further optimizes the model’s performance in legal text classification tasks.