EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Query-Driven Graph Retrieval Framework with Adaptive Pruning for Multi-Hop Question Answering

Hao Wang, Tianyue Wang, Zhongrui Sun, He Li, Zhengyang Cao, Lihang Feng, Dong Wang

Electronics · 18.03.2026 · 15 (6) · 1263

DOI: 10.3390/electronics15061263 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Electronics, online erschienen am 18.03.2026. Verfasst von Hao Wang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Jiangsu Science and Technology Programme.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Multi-hop question answering (MHQA) requires models to retrieve and reason over evidence distributed across multiple documents, which remains challenging for conventional retrieval-augmented generation (RAG) approaches. Although RAG improves factual grounding by incorporating external knowledge, flat retrieval strategies often struggle to maintain coherent reasoning chains when implicit dependencies among entities and documents are involved. This paper presents a query-driven dual-layer graph retrieval framework for MHQA. The framework operates on a unified heterogeneous graph integrating entities, relations, and supporting texts, and dynamically constructs candidate subgraphs through joint retrieval over entities and relations, complemented by lexical retrieval signals. Reasoning paths are refined by combining structural strength modeling with contrastive learning-based path scoring, and an adaptive pruning strategy is employed to regulate evidence scale according to query complexity and path score distributions. Experiments on HotpotQA and 2WikiMultihopQA show that the proposed framework achieves higher EM and F1 scores than existing RAG and graph-based retrieval methods, particularly in complex multi-hop scenarios. These results indicate the importance of structured and query-adaptive evidence organization for multi-hop reasoning.

Förderung

  • National Natural Science Foundation of China
  • Jiangsu Science and Technology Programme

Zitieren

APA

Wang, H., Wang, T., Sun, Z., Li, H., Cao, Z., Feng, L., & Wang, D. (2026). A Query-Driven Graph Retrieval Framework with Adaptive Pruning for Multi-Hop Question Answering. Electronics, 15(6), 1263. https://doi.org/10.3390/electronics15061263

Vancouver

Wang H, Wang T, Sun Z, Li H, Cao Z, Feng L, et al. A Query-Driven Graph Retrieval Framework with Adaptive Pruning for Multi-Hop Question Answering. Electronics. 2026;15(6):1263. doi:10.3390/electronics15061263

BibTeX
@article{Wang2026,
  author = {Wang, Hao and Wang, Tianyue and Sun, Zhongrui and Li, He and Cao, Zhengyang and Feng, Lihang and Wang, Dong},
  title = {A Query-Driven Graph Retrieval Framework with Adaptive Pruning for Multi-Hop Question Answering},
  journal = {Electronics},
  year = {2026},
  volume = {15},
  number = {6},
  pages = {1263},
  doi = {10.3390/electronics15061263}
}
Kostenlos starten →