LRAS: Advanced Legal Reasoning with Agentic Search

TL;DR

LRAS introduces an interactive framework for legal reasoning that enhances large models' ability to identify knowledge gaps and improve reasoning accuracy through active inquiry, outperforming existing methods.

Contribution

It is the first framework to enable legal LLMs to switch from static reasoning to dynamic, interactive inquiry using introspective and reinforcement learning techniques.

Findings

LRAS outperforms baselines by 8.2-32% in legal reasoning tasks.

It effectively identifies knowledge boundaries and handles reasoning complexity.

Empirical results show significant improvements in deep legal reasoning accuracy.

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.