Search papers, labs, and topics across Lattice.
This paper introduces DeepRepoQA, a novel framework for answering developer questions about software repositories that overcomes the limitations of existing methods by enabling deep reasoning across multiple files and complex architectures. Utilizing a Monte-Carlo Tree Search (MCTS) mechanism, DeepRepoQA allows LLM agents to systematically explore and navigate repository structures, facilitating effective multi-hop reasoning over long-range code dependencies. Experimental results on the SWE-QA benchmark show significant performance improvements compared to strong baselines, highlighting the framework's capability for enhanced repository-level understanding.
DeepRepoQA achieves substantial performance gains in repository question answering by leveraging Monte-Carlo Tree Search for deep, multi-hop reasoning over code dependencies.
Answering developer questions about a software repository is a critical yet under-explored problem in software engineering. While existing repository understanding methods have advanced the field, they predominantly rely on surface-level code retrieval and lack the ability for deep reasoning over multiple files, complex software architectures, and grounding answers in long-range code dependencies. To address these limitations, we propose DeepRepoQA, a novel question answering (QA) framework for repository-level code understanding. DeepRepoQA builds on an agentic framework where LLM agents find answers through a systematic tree search over the repository structure. A Monte-Carlo Tree Search (MCTS) mechanism is employed to empower agents to dynamically search, navigate, and inspect code, enabling effective multi-hop reasoning over long-range code dependencies. Comprehensive experiments on the SWE-QA benchmark demonstrate substantial performance gains over strong baselines, validating the effectiveness of systematic MCTS-guided exploration for multi-hop repository reasoning.
