RTI-Bench: A Structured Dataset for Indian Right-to-Information Decision Analysis

2026-05-19 · Source: cs.CL updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Advanced, quick

Summary

RTI-Bench is a newly introduced structured dataset designed to aid in the analysis of Indian Right to Information (RTI) Act, 2005 decisions. This dataset addresses the challenge citizens face in understanding complex administrative language and predicting appeal outcomes from Central Information Commission (CIC) decisions. It includes outcome labels, exemption citations, IRAC-style reasoning components, and procedural timelines. RTI-Bench integrates 1,218 cases from an existing instruction-response corpus, enhanced with rule-based extraction, and 298 CIC decision PDFs collected directly from the Commission portal, spanning decisions from 2023 to 2026 across five commissioners and three document formats. The dataset achieves 89% label coverage on the instruction-response corpus and 51% on the PDF subset's 239 primary decisions, with a manual review yielding 95.3% label precision on a 50-case sample. A zero-shot Mistral 7B baseline demonstrated 57.3% accuracy and 37.0% macro-F1 for outcome prediction on 100 cases, significantly outperforming the 14.3% majority-class baseline.

Key takeaway

For NLP Engineers developing legal AI tools for India, RTI-Bench offers a critical resource for training and evaluating models on administrative decision analysis. Your work can now leverage this structured dataset to build systems that predict appeal outcomes or clarify complex legal language, potentially improving citizen access to information. Consider integrating RTI-Bench to enhance the accuracy and relevance of your legal tech applications.

Key insights

RTI-Bench provides the first structured dataset for Indian RTI decisions, enabling AI-driven analysis of complex legal texts.

Principles

• Structured data improves legal decision analysis.
• Rule-based extraction enhances existing corpora.

Method

The dataset combines rule-based extraction from an instruction-response corpus with direct PDF collection, followed by manual review for precision and baseline model evaluation.

In practice

• Use RTI-Bench for legal outcome prediction.
• Apply IRAC-style reasoning to administrative texts.

Topics

• RTI-Bench Dataset
• Indian Right to Information Act
• Central Information Commission
• Legal Decision Analysis
• Natural Language Processing

Best for: AI Scientist, Research Scientist, NLP Engineer

Related on AIssential

• VLegal-Bench: Cognitively Grounded Benchmark for Vietnamese Legal Reasoning of Large Language Models — VLegal-Bench offers a cognitively-grounded benchmark for LLMs in Vietnamese civil law, revealing specialized models outperform general ones on complex legal reasoning.
• After trial judge cites AI-generated judgment, Supreme Court asks BCI to set up expert panel to examine issue - Bar and Bench — AI-generated legal "hallucinations" necessitate judicial accountability and sovereign AI development in India.
• IndiaFinBench: An Evaluation Benchmark for Large Language Model Performance on Indian Financial Regulatory Text — IndiaFinBench provides the first public benchmark for LLMs on Indian financial regulatory text.
• Why Native Legal AI is Required for India — Native Legal AI, tailored to specific national legal systems, overcomes limitations of generic AI in complex jurisdictions.
• LAUKIN: A Multi-jurisdictional Common Law Contract Dataset — LAUKIN provides a multi-jurisdictional contract dataset revealing significant legal drafting divergences, challenging automated equivalence classification.
• HKJudge: A Legal Discourse-Annotated Corpus for Interpreting What Courts Find, How They Reason, and What They Rule — Expert-annotated legal discourse corpora enable advanced AI understanding of court judgments.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CL updates on arXiv.org.