LeKUBE: A Knowledge Update BEnchmark for Legal Domain

• DOI: 10.1145/3673791.3698407
• Publication Date: 2024-12-08T11:24:16Z
• Authors: Changyue Wang; Weihang Su; Yiran Hu; Qingyao Ai; Yueyue Wu; Cheng Luo; Yiqun Liu; Min Zhang; Shaoping Ma
• Keywords: Benchmark (surveying)

Recent advances in Large Language Models (LLMs) have significantly shaped the applications of AI multiple fields, including studies legal intelligence. Trained on extensive texts, statutes and documents, LLMs can capture important knowledge/concepts effectively provide support for downstream such as consultancy. Yet, dynamic nature interpretations also poses new challenges to use applications. Particularly, how update knowledge efficiently has become an research problem practice. Existing benchmarks evaluating methods are mostly designed open domain cannot address specific domain, nuanced application knowledge, complexity lengthiness regulations, intricate reasoning.