Reproducibility Package for Benchmarking Long-Context LLMs and RAG Baselines for Corpus-Based Chinese-English Linguistic Analysis

This record provides the reproducibility package for the study "Benchmarking Long-Context LLMs and Retrieval-Augmented Generation Baselines for Corpus-Based Chinese-English Linguistic Analysis." It includes formal gold labels and blind item metadata, model request templates, parsing and evaluation scripts, corrected output files, bootstrap confidence intervals, confusion summaries, figure-generation scripts, and documentation needed to reproduce the reported LC direct, BM25 RAG, hybrid RAG, and dense RAG evaluations. Redistribution of OPUS-derived materials is limited according to the original corpus licenses; where full redistribution is not permitted, reconstruction scripts or item identifiers are provided.