Work

Deka Retrieval: Component-Level Criteria for Mining Legal Case Pairs in Thai Supreme Court Decisions

Year: 2026

Type: conference-paper

Abstract: Legal case retrieval (LCR) defines relevance beyond text similarity, judgments may describe near-identical events yet be decided under different charges.Existing synthesis pipelines mine training pairs from text similarity and shared charges, which separates positives from negatives but leaves the difficulty of each pair implicit.We present Deka Retrieval, an LCR benchmark on Thai Supreme Court decisions with an LLM relevance cases mining pipeline for a civil-law jurisdiction.We decompose each judgment into case components (Charge, Fact, Reasoning, Decision) and split easy/hard level on exist binary signals and introduce an anomaly variant a conflict judgment that shares the anchor's charge and facts but reaching a different reasoning and decision.Retriever performance follows query length and legal terminology: Fact > Lawyer expert > Lay people.Fine-tuning gains most on the weakest type, lifting Lay R@10 by 2.6 × (7.6 → 19.5) versus 1.5× on Lawyer and 1.3× on Fact.Training on the shortest query also transfers to the longer ones, but not the reverse.Adding anomaly cases in training as positive degrades the overall performance, compare with without it.Our dataset and models are available at https://huggingface.co/collections/ VISAI-AI/deka-retrieval

Source: Natural Legal Language Processing Workshop (NLLP)

DOI: 10.18653/2026.nllp-1.35

Authors: Pasin Buakhaw, Pawitsapak Akarajaradwong, Chompakorn Chaksangchaichot

Institution: Chulalongkorn University

Language: English

Topic: Artificial Intelligence in Law

Subfield: Political Science and International Relations

Field: Social Sciences

Domain: Social Sciences

SDG: Peace, justice, and strong institutions

Open Access status: bronze

Funder: Vidyasirimedhi Institute of Science and Technology