S-Gens: Structure-Aware Synthetic Data Generation for Enhancing Reasoning-Intensive Dense Retrieval

Dense retrievers rely heavily on high-quality training triplets, yet existing data construction strategies remain inadequate for reasoning-intensive retrieval tasks involving multi-hop reasoning, entity relation tracing, and implicit evidence composition. Positive samples are often based on shallow...

ver descrição completa

Detalhes bibliográficos

Principais autores:	Zhou Lei, Yanqi Xu, Shengbo Chen
Formato:	Artigo
Idioma:	Inglês
Publicado em:	MDPI AG 2026-04-01
Colecção:	Information
Assuntos:	dense retrieval synthetic data generation knowledge graph hard negatives reasoning-enhanced retrieval
Acesso em linha:	https://www.mdpi.com/2078-2489/17/5/413
Tags:	Adicionar Tag Sem tags, seja o primeiro a adicionar uma tag!