DEO: Training-Free Direct Embedding Optimization for Negation-Aware Retrieval

TL;DR

DEO is a training-free method that enhances negation-aware retrieval by optimizing query embeddings through contrastive objectives, outperforming existing methods without additional training or fine-tuning.

Contribution

DEO introduces a novel, training-free approach for negation-aware retrieval that decomposes queries and optimizes embeddings, avoiding extra training data or model updates.

Findings

DEO improves nDCG@10 by +0.0738 on NegConstraint.

DEO increases MAP@100 by +0.1028 on NegConstraint.

DEO enhances Recall@5 by +6% in multimodal retrieval.

Abstract

Recent advances in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) have enabled diverse retrieval methods. However, existing retrieval methods often fail to accurately retrieve results for negation and exclusion queries. To address this limitation, prior approaches rely on embedding adaptation or fine-tuning, which introduce additional computational cost and deployment complexity. We propose Direct Embedding Optimization (DEO), a training-free method for negation-aware text and multimodal retrieval. DEO decomposes queries into positive and negative components and optimizes the query embedding with a contrastive objective. Without additional training data or model updates, DEO outperforms baselines on NegConstraint, with gains of +0.0738 nDCG@10 and +0.1028 MAP@100, while improving Recall@5 by +6\% over OpenAI CLIP in multimodal retrieval. These results demonstrate the practicality of DEO for negation- and exclusion-aware retrieval in real-world settings.