NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning

2026-07-17

AI Summary

This work addresses five core tasks in legal information retrieval and reasoning—case retrieval, statute retrieval, textual entailment, entailment judgment, and judgment prediction—by proposing a multi-stage adaptive framework that dynamically integrates filtering, dense retrieval, reranking, and large language model–based reasoning. The approach innovatively incorporates query-adaptive truncation, dynamic routing solvers, consensus-based ensemble verification, and argumentation-graph–guided probabilistic inference. It further leverages complementary embeddings, a fine-tuned generative reranker, a hierarchical Transformer-CRF architecture, and zero-shot chain-of-thought prompting. Evaluated on all tasks of the COLIEE 2026 benchmark, the method achieves state-of-the-art performance, significantly enhancing accuracy, adaptability, and generalization in complex legal scenarios.

Abstract

This paper presents the methodologies and results of the NOWJ team's participation across all five tasks of the COLIEE 2026 competition. For Task 1 (Legal Case Retrieval), we propose a four-stage pipeline comprising candidate filtering, dense retrieval with complementary embedding models, cross-encoder reranking via fine-tuned generative rerankers and MLP-based pairwise classification, and adaptive per-query cutoff prediction. For Task 2 (Legal Case Entailment), we combine BM25 filtering, T5-based reranking, and LLM-based entailment verification with consensus ensemble. For Task 3 (Statute Law Retrieval and Entailment), we adopt a retrieval-augmented generation framework with dense retrieval, attention-based reranking, and few-shot-prompted LLM reasoning. For Task 4 (Legal Textual Entailment), we introduce a dynamic routing pipeline that classifies query difficulty and dispatches cases to either a balanced few-shot solver or a structured zero-shot chain-of-thought solver. For the Pilot Task (Legal Judgment Prediction), we combine hierarchical transformers with CRF layers, argument relation mining, and probabilistic argumentation graph reasoning.

Problem

• Legal Retrieval
• Legal Reasoning
• Textual Entailment
• Judgment Prediction
• Statute Law

Innovation

• adaptive pipeline
• legal retrieval
• generative reranker
• dynamic routing
• retrieval-augmented reasoning