KCC: Korean Civil Case Dataset for Legal Information Retrieval

Anonymous

TL;DR: We introduce the KCC dataset with 2,942 Korean civil cases (1947-2022), featuring a four-level case similarity criteria by legal experts, aimed at improving legal information retrieval and aiding professionals.

Abstract: Analyzing relevant or similar precedent cases is crucial in the field of law. This study presents a novel legal information retrieval dataset, KCC, for Korean civil judgments, consisting of 2,942 civil cases, treated by Korean courts in between March 3rd in 1947 and December 31st, 2022. In the proposed dataset, we introduce and annotate a 4-level case similarity criteria, which is verified by legal experts, resulting in both high-level legal reasoning as well as factual circumstances can be considered in the legal IR tasks. Experiments on the proposed dataset using popular legal IR methods demonstrate promising performance in legal IR tasks. We believe the proposed dataset can be used as valuable resource for developing legal IR models, which can assist legal professionals.

Paper Type: short

Research Area: Resources and Evaluation

Contribution Types: Data resources

Languages Studied: Korean