Request for the Full 14,049-Case Corpus and Construction Scripts Used in Table 2 更简洁一点也可以：

[liuliuliuyyy]

The paper states that the retrieval corpus contains 14,049 cases aggregated from JuDGE, CAIL2018, CMDL, and LeCaRDv2. However, the public data-preparation pipeline only calls prepare_cail_data.py, and the provided cases_with_feature.json appears to be a small CAIL example. Could you provide the corpus construction script, source case IDs, or the processed 14,049-case corpus used for Table 2?

[rzchen777-wq]

Thank you for your interest in our work. The original datas/cases_with_feature.json is a small CAIL demonstration corpus rather than the retrieval corpus used for Table 2.

The repository has been updated with the relevant materials:

• The complete 14,049-case corpus: datas/main_experiment/cases_with_feature_big.json
• The processed case ID ranges and primary source splits: datas/main_experiment/corpus_sources.json
• The sampling and corpus-merging scripts: scripts/sample_source_cases.py and scripts/merge_retrieval_corpus.py
• Construction details and example commands: docs/TABLE2_REPRODUCTION.md

LeCaRDv2 was used during manual corpus validation, and a small amount of supplementary data was incorporated without separate source-ID tags. The released 14,049-case file is the processed corpus used for the Table 2 experiments.