Backend
NONSENSE: Non-Semantic Strings Expose Semanticity Gaps in Dense Retrieval
arXiv cs.IR0 views
arXiv:2611.22353v1 Announce Type: new Abstract: Dense retrieval is central to modern search, question answering, and retrieval-augmented generation, where high similarity is often treated as evidence of semantic relevance. We examine how far this assumption can be stressed. NONSENSE is a black-box sparse-edit probe that constructs visibly non-semantic strings while forbidding query-term copying, optimizing only the dense similarity between a query and the string. Across five BEIR-style datasets and three popular single-vector retrievers, these strings frequently outscore the retriever's own best judged query-relevant passage (QRP), and often rank first in the full corpus. The result indicates a semanticity gap in the evaluated dense retrievers: nearest-neighbor status can be separated from passage-like meaning. These findings highlight the importance of evaluating dense retrievers not only by whether they retrieve relevant documents, but also by how reliably their similarity scores reflect semantic evidence.
Source:https://arxiv.org/abs/2611.22353PreviousTeaching Language Models to Narrow Boolean Patent Searches from Search FeedbackNextDeal With It: On Better Schedules for Listwise Reranking
Related posts
FastAPI 0.163's new dependency concurrency limit deadlocked two of my routes
0dev.to
AI-Generated Code Can Look Correct and Still Be Wrong
0dev.to
How Website Speed Affects SEO and User Experience
0dev.to
Building a Client Portal on the Retell API: The Eight Line Items Nobody Estimates
0dev.to
Why Azure for SAP Workloads Is Becoming an Important Skill for Cloud Engineers
0dev.to
Lessons from Building a Privacy-First AST & Hot-Path Code Review Assistant in Rust (and Supporting Open Source)
0dev.to