Factuality and Transparency Are All RAG Needs! Self-Explaining Contrastive Evidence Re-ranking

TL;DR

This paper presents CER, a novel retrieval re-ranking method that enhances factuality and transparency by fine-tuning embeddings with contrastive learning and generating rationales, improving accuracy and reliability in safety-critical applications.

Contribution

CER introduces a contrastive learning-based re-ranking approach that aligns embeddings with evidential reasoning and generates token-level rationales for improved transparency.

Findings

Improves retrieval accuracy in clinical trial reports.

Reduces hallucinations in RAG systems.

Provides transparent, evidence-based retrieval.

Abstract

This extended abstract introduces Self-Explaining Contrastive Evidence Re-Ranking (CER), a novel method that restructures retrieval around factual evidence by fine-tuning embeddings with contrastive learning and generating token-level attribution rationales for each retrieved passage. Hard negatives are automatically selected using a subjectivity-based criterion, forcing the model to pull factual rationales closer while pushing subjective or misleading explanations apart. As a result, the method creates an embedding space explicitly aligned with evidential reasoning. We evaluated our method on clinical trial reports, and initial experimental results show that CER improves retrieval accuracy, mitigates the potential for hallucinations in RAG systems, and provides transparent, evidence-based retrieval that enhances reliability, especially in safety-critical domains.