Autonomous Knowledge Graph Exploration with Adaptive Breadth-Depth Retrieval

TL;DR

ARK is a flexible, training-free knowledge graph retrieval method that dynamically balances broad and deep search strategies, significantly improving evidence retrieval for language models.

Contribution

Introduces ARK, a novel adaptive retrieval approach that combines global lexical search and neighborhood exploration without training, enhancing knowledge graph evidence retrieval.

Findings

ARK achieves 59.1% Hit@1 and 67.4 MRR on STaRK, outperforming existing methods.

ARK's trajectories distilled into an 8B model improve accuracy on AMAZON, MAG, and PRIME datasets.

The method adapts to query types, using global search for language-heavy and neighborhood exploration for relation-heavy queries.

Abstract

Retrieving evidence for language model queries from knowledge graphs requires balancing broad search across the graph with multi-hop traversal to follow relational links. Similarity-based retrievers provide coverage but remain shallow, whereas traversal-based methods rely on selecting seed nodes to start exploration, which can fail when queries span multiple entities and relations. We introduce ARK: Adaptive Retriever of Knowledge, a tool-using KG retriever that gives a language model control over this breadth-depth tradeoff using a two-operation toolset: global lexical search over node descriptors and one-hop neighborhood exploration that composes into multi-hop traversal. ARK alternates between breadth-oriented discovery and depth-oriented expansion without depending on a fragile seed selection, a pre-set hop depth, or requiring retrieval training. ARK adapts tool use to queries, using global search for language-heavy queries and neighborhood exploration for relation-heavy queries. On STaRK, ARK reaches 59.1% average Hit@1 and 67.4 average MRR, improving average Hit@1 by up to 31.4% and average MRR by up to 28.0% over retrieval-based and agent-based training-free methods. Finally, we distill ARK's tool-use trajectories from a large teacher into an 8B model via label-free imitation, improving Hit@1 by +7.0, +26.6, and +13.5 absolute points over the base 8B model on AMAZON, MAG, and PRIME datasets, respectively, while retaining up to 98.5% of the teacher's Hit@1 rate.