Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Group-Aware Adaptive Retrieval for Evidence Navigation (GAREN), a novel approach that enhances reasoning-intensive retrieval by organizing documents into semantically coherent groups within a corpus graph. By leveraging group-level expansion rather than traditional document-level isolation, GAREN effectively addresses the bounded recall problem, leading to more relevant document retrieval. Experimental results demonstrate that GAREN achieves up to an 8.0% improvement over existing state-of-the-art methods on the BRIGHT benchmark.
Group-level navigation in document retrieval can boost relevance by up to 8% compared to traditional methods.
Reasoning-intensive retrieval addresses queries whose relevance cannot be identified by surface-level matching, thereby requiring multi-step reasoning. Because relevant documents rarely appear in the initial candidate set, retrieval systems suffer from the bounded recall problem. Existing methods iteratively expand a candidate pool at the document level over a corpus graph, examining each neighbor in isolation and drifting toward a narrow region of the corpus. To address this problem, we propose Group-Aware Adaptive Retrieval for Evidence Navigation (GAREN), which explores the corpus graph through group-level expansion. GAREN organizes documents into semantically coherent and distinguishable groups based on their connections in the corpus graph. The information in each group indicates what can be accessed by expanding through it, providing guidance beyond individual document-level signals. At each iteration, GAREN uses a group-level navigator to identify promising expansion directions, retrieves documents from the selected groups, and applies a document-level reranker to evaluate the updated candidate set. Extensive experiments show that GAREN achieves up to 8.0% improvement over the strongest baseline on BRIGHT. The source code is available at https://github.com/KJ12124/GAREN