Mixture of Structural-and-Textual Retrieval over Text-rich Graph Knowledge Bases
Yongjia Lei, Haoyu Han, Ryan A. Rossi, Franck Dernoncourt, Nedim Lipka, Mahantesh M Halappanavar, Jiliang Tang, Yu Wang
2025-03-06
Summary
This paper talks about a new method called MoR (Mixture of Structural-and-Textual Retrieval) that helps AI systems find information more effectively in text-rich graph knowledge bases, which are like super-smart digital libraries
What's the problem?
Current methods for finding information in these knowledge bases often look at text and structure separately, which means they miss out on important connections. Some methods even ignore the structure completely after a certain point, which can lead to less accurate results
What's the solution?
The researchers created MoR, which uses a three-step process: Planning, Reasoning, and Organizing. In Planning, it creates a map of how to answer the question. In Reasoning, it looks at both the text and the structure of the information to find good answers. In Organizing, it ranks the answers based on how they fit into the overall structure of the knowledge base
Why it matters?
This matters because it helps AI systems give more accurate and relevant answers to complex questions. By considering both the text and the structure of information, MoR can find connections that other methods might miss. This could lead to better search engines, more helpful AI assistants, and improved tools for researchers and students looking for information in large databases
Abstract
Text-rich Graph Knowledge Bases (TG-KBs) have become increasingly crucial for answering queries by providing textual and structural knowledge. However, current retrieval methods often retrieve these two types of knowledge in isolation without considering their mutual reinforcement and some hybrid methods even bypass structural retrieval entirely after neighboring aggregation. To fill in this gap, we propose a Mixture of Structural-and-Textual Retrieval (MoR) to retrieve these two types of knowledge via a Planning-Reasoning-Organizing framework. In the Planning stage, MoR generates textual planning graphs delineating the logic for answering queries. Following planning graphs, in the Reasoning stage, MoR interweaves structural traversal and textual matching to obtain candidates from TG-KBs. In the Organizing stage, MoR further reranks fetched candidates based on their structural trajectory. Extensive experiments demonstrate the superiority of MoR in harmonizing structural and textual retrieval with insights, including uneven retrieving performance across different query logics and the benefits of integrating structural trajectories for candidate reranking. Our code is available at https://github.com/Yoega/MoR.