E-SENS: Exclusion-Sensitive Penalization for Negative-Constraint Retrieval
2026-08-31 • Information Retrieval
Information RetrievalArtificial Intelligence
AI summaryⓘ
The authors address a problem where language models that use search can mistakenly show results about things users want to avoid, especially when the unwanted concept appears in the search query. They propose a method called E-SENS, which adjusts how results are ranked by identifying and subtracting similarity scores for the unwanted concepts. This method does not need additional training and helps reduce unwanted results while keeping the relevant ones. Their tests show E-SENS improves retrieval when users want to exclude certain topics.
retrieval-augmented language modelsnegative constraintsdense retrieversrerankingnegation sensitivityquery embeddingExcluIR datasetrecall-violation trade-offtrap queryembedding models
Authors
Yerang Kim, Jiyoon Myung, Joohyung Han
Abstract
Retrieval-augmented language models can fail to respect negative constraints when the retriever supplies evidence about concepts the user explicitly excluded. Beyond explicit negation, queries may ask for answers that include one concept while excluding another, or for entities that belong to a category but differ from a closely related instance. Because the excluded concept still appears in the query text, dense retrievers may assign high similarity to documents about that concept even when the user asks to avoid it. We introduce E-SENS, a training-free reranking method for negation-sensitive retrieval. E-SENS extracts a compact trap query for the excluded side and subtracts trap-query similarity from the original-query retrieval score. On ExcluIR, E-SENS shows a clear recall-violation trade-off across four embedding models and reduces trap retrieval at recall-preserving settings.