PropRAG: Guiding Retrieval with Beam Search over Proposition Paths

ACL ARR 2025 May Submission4228 Authors

Abstract: Retrieval Augmented Generation (RAG) has become the standard non-parametric approach for equipping Large Language Models (LLMs) with up-to-date knowledge and mitigating catastrophic forgetting common in continual learning. However, standard RAG, relying on independent passage retrieval, fails to capture the interconnected nature of human memory crucial for complex reasoning (associativity) and contextual understanding (sense-making). While structured RAG methods like HippoRAG 2 utilize knowledge graphs built from triples, we argue that the inherent context loss of knowledge triples limits fidelity. We introduce PropRAG, leveraging context-rich propositions and a novel LLM-free online beam search over proposition paths to find multi-step reasoning chains. PropRAG achieves state-of-the-art zero-shot Recall@5 and F1 scores on 2Wiki, HotpotQA, and MuSiQue, advancing non-parametric continual learning by improving evidence retrieval through richer representation and efficient reasoning path discovery.

Paper Type: Long

Research Area: Language Modeling

Research Area Keywords: retrieval-augmented generation, passage retrieval, graph-based methods

Contribution Types: Model analysis & interpretability, NLP engineering experiment

Languages Studied: English

Submission Number: 4228