Search papers, labs, and topics across Lattice.
The paper introduces SEARCH-R, a framework for multi-hop question answering that combines a fine-tuned Llama3.1-8B model for structured reasoning path generation with a dependency tree-based retrieval method. SEARCH-R addresses limitations in existing MHQA approaches by training an end-to-end reasoning path navigator for sub-question decomposition and evaluating the informational contribution of documents using dependency trees. Experiments on multi-hop datasets demonstrate the effectiveness of SEARCH-R in generating correct reasoning paths and retrieving useful knowledge.
Stop relying on LLMs to "hallucinate" reasoning paths – SEARCH-R uses a fine-tuned Llama3.1-8B model and dependency tree-based retrieval to navigate multi-hop question answering more reliably.
Multi-hop Question Answering (MHQA) aims to answer questions that require multi-step reasoning. It presents two key challenges: generating correct reasoning paths in response to the complex user queries, and accurately retrieving essential knowledge in the face of potential limitations in large language models (LLMs). Existing approaches primarily rely on prompt-based methods to generate reasoning paths, which are further combined with traditional sparse or dense retrieval to produce the final answer. However, the generation of reasoning paths commonly lacks effective control over the generative process, thus leading the reasoning astray. Meanwhile, the retrieval methods over-rely on knowledge matching or similarity scores rather than evaluating the practical utility of the information, resulting in retrieving homogeneous or non-useful information. Therefore, we propose a Structured Entity-Aware Retrieval with Chain-of-Reasoning Navigator framework named SEARCH-R. Specifically, SEARCH-R trains an end-to-end reasoning path navigator, which is able to provide a powerful sub-question decomposer by fine-tuning the Llama3.1-8B model. Moreover, a novel dependency tree-based retrieval is designed to evaluate the informational contribution of the document quantitatively. Extensive experiments on three challenging multi-hop datasets validate the effectiveness of the proposed framework. The code and dataset are available at: https://github.com/Applied-Machine-Learning-Lab/ACL2026_SEARCH-R.
