Evaluation Artefacts for LLM@PR: A LLM-driven Petition Ranking Framework

This repository contains evaluation artefacts supporting the paper:"LLMPR: A Novel LLM-Driven Transfer Learning based Petition Ranking Model" (submitted to Artificial Intelligence and Law, 2025). The files include:- Test-set predictions for all models used (LightGBM regressors with embeddings and numeric features)- Bootstrapped 95% confidence intervals for Tolerance-10% accuracy- Duplicate-content validation metrics- A metrics summary table across models All results are based on the ILDC dataset with its official train/test split. The code used for this evaluation is compatible with Python 3.8+ and scikit-learn, LightGBM, numpy, pandas. See `README_LLMPR_Artefacts.md` for details and reproduction instructions.