DESA：混合检索中的通道非对称查询扩展方法
原标题：Dense Expands, Sparse Anchors: Channel-Asymmetric Query Expansion for Hybrid Retrieval
AI 摘要
该研究提出 DESA（Dense Expansion and Sparse Anchoring）方法，用于混合检索中的非对称查询扩展。LLM 生成参考段落，通过正交残差扩展增强稠密检索，通过分数乘积锚定限制稀疏检索的词汇支持。在七个 BEIR 数据集上，DESA 相比未扩展查询将 nDCG@10 和 Recall@20 分别提升 3.82% 和 2.38%，同时将稠密和稀疏访问深度分别降低 36.90% 和 36.56%。研究强调在完整列表融合下评估检索效果与访问深度，并指出扩展效果依赖于检索器。
正文节选
Dense Expands, Sparse Anchors: Channel-Asymmetric Query Expansion for Hybrid Retrieval Abstract LLM-based query expansion improves retrieval by generating document-like passages. In hybrid retrieval, however, most evaluations fuse fixed top- dense and sparse rankings. Because the cutoff controls both which cross-channel contributions enter fusion and how much of each ranking is accessed, gains measured at one can change or reverse at another. We separate these effects by evaluating retrieval eff