A Navigational Approach for Comprehensive RAG via Traversal over Proposition Graphs
Maxime Delmas, Lei Xu, Andr\'e Freitas

TL;DR
The paper introduces ToPG, a novel retrieval-augmented generation framework that models knowledge as a heterogeneous proposition graph, enabling effective multi-hop and fact-oriented queries through query-aware graph traversal and iterative refinement.
Contribution
ToPG is the first framework to combine proposition graphs with query-aware traversal and LLM feedback for comprehensive RAG performance.
Findings
ToPG outperforms existing methods on multiple QA tasks.
Effective multi-hop reasoning achieved through graph traversal.
Improved fact retrieval with structured proposition graphs.
Abstract
Standard RAG pipelines based on chunking excel at simple factual retrieval but fail on complex multi-hop queries due to a lack of structural connectivity. Conversely, initial strategies that interleave retrieval with reasoning often lack global corpus awareness, while Knowledge Graph (KG)-based RAG performs strongly on complex multi-hop tasks but suffers on fact-oriented single-hop queries. To bridge this gap, we propose a novel RAG framework: ToPG (Traversal over Proposition Graphs). ToPG models its knowledge base as a heterogeneous graph of propositions, entities, and passages, effectively combining the granular fact density of propositions with graph connectivity. We leverage this structure using iterative Suggestion-Selection cycles, where the Suggestion phase enables a query-aware traversal of the graph, and the Selection phase provides LLM feedback to prune irrelevant propositions…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Graph Neural Networks · Topic Modeling · Multimodal Machine Learning Applications
