LRAS: Advanced Legal Reasoning with Agentic Search

2026-01-12

AI Summary

This work addresses a critical limitation of current legal large language models, which rely on closed-world reasoning and lack awareness of their knowledge boundaries, often producing confidently incorrect conclusions. To overcome this, we propose the LRAS framework—the first to integrate an active querying mechanism into legal reasoning—enabling dynamic interaction between knowledge boundary recognition and complex inference. LRAS combines introspective imitation learning with difficulty-aware reinforcement learning, synergistically coupling agent-based active retrieval with legal large language models. This approach significantly enhances reasoning rigor, outperforming state-of-the-art methods by 8.2%–32% on challenging legal reasoning benchmarks, with particularly pronounced gains in tasks requiring reliable external legal knowledge.

Technology Category

• Natural Language Processing: Question Answering
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Knowledge Representation and Reasoning: Argumentation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

While Large Reasoning Models (LRMs) have demonstrated exceptional logical capabilities in mathematical domains, their application to the legal field remains hindered by the strict requirements for procedural rigor and adherence to legal logic. Existing legal LLMs, which rely on"closed-loop reasoning"derived solely from internal parametric knowledge, frequently suffer from lack of self-awareness regarding their knowledge boundaries, leading to confident yet incorrect conclusions. To address this challenge, we present Legal Reasoning with Agentic Search (LRAS), the first framework designed to transition legal LLMs from static and parametric"closed-loop thinking"to dynamic and interactive"Active Inquiry". By integrating Introspective Imitation Learning and Difficulty-aware Reinforcement Learning, LRAS enables LRMs to identify knowledge boundaries and handle legal reasoning complexity. Empirical results demonstrate that LRAS outperforms state-of-the-art baselines by 8.2-32\%, with the most substantial gains observed in tasks requiring deep reasoning with reliable knowledge. We will release our data and models for further exploration soon.

Problem

• Legal Reasoning
• Closed-loop Reasoning
• Knowledge Boundaries
• Procedural Rigor
• Legal Logic

Innovation

• Agentic Search
• Legal Reasoning
• Introspective Imitation Learning
• Difficulty-aware Reinforcement Learning
• Active Inquiry