P Moth-Retrieval: Graph-Free Multi-Hop Retrieval via Query-Time Orchestration (Beating Graph-Based Systems on HotpotQA) [P]
Summary
Moth-Retrieval presents a graph-free approach to multi-hop retrieval that outperforms graph-based systems on the HotpotQA benchmark.
Similar Articles
Most multi-hop RAG goes stale the moment your data changes, what about a training-free approach that skips the graph rebuild?
Presents a training-free method for multi-hop retrieval-augmented generation that avoids costly graph rebuilds when underlying data changes, tackling the staleness issue in dynamic environments.
We open-sourced a graph-free multi-hop RAG framework — matches Graph-RAG accuracy without the rebuild cost (Apache-2.0)
MOTHRAG is a graph-free multi-hop RAG framework that matches the accuracy of graph-based systems like GraphRAG and HippoRAG on benchmarks, while avoiding costly graph rebuilds by using a dense index and query-time orchestration.
EviReform: Evidence-Guided Query Reformulation for Multi-Hop Graph Retrieval
EviReform 提出了一种证据引导的查询重写方法，用于多跳图检索，将检索请求的修订与证据聚合分离，并在 2WikiMultiHopQA、HotpotQA 和 MuSiQue 上超过了最强基线，分别提升了最多 5.59 Recall@5 和 4.50 F1。
@garrytan: Lots of retrieval systems seem to give you keyword matching and graph retrieval in a raw format This plus graph travers…
Garry Tan highlights a retrieval system that uniquely combines keyword matching, graph traversal, and gap analysis, an approach not seen elsewhere.
Co-Evolving Graph and Text Memory for Training-Free Multi-Hop Question Answering
Proposes Co-E, a training-free system that synchronizes graph and text memory for multi-hop question answering, improving over comparable training-free baselines on six benchmarks.