eTracer: Towards Traceable Text Generation via Claim-Level Grounding

TL;DR

eTracer is a framework for traceable text generation that aligns claims with evidence to improve verification and trustworthiness, especially in high-stakes biomedical applications.

Contribution

It introduces claim-level grounding for verifiable responses, enhancing alignment accuracy and enabling precise source tracing and faithfulness quantification.

Findings

Improved grounding quality over conventional methods

Enhanced verification efficiency for users

Effective claim-evidence alignment in biomedical texts

Abstract

How can system-generated responses be efficiently verified, especially in the high-stakes biomedical domain? To address this challenge, we introduce eTracer, a plug-and-play framework that enables traceable text generation by grounding claims against contextual evidence. Through post-hoc grounding, each response claim is aligned with contextual evidence that either supports or contradicts it. Building on claim-level grounding results, eTracer not only enables users to precisely trace responses back to their contextual source but also quantifies response faithfulness, thereby enabling the verifiability and trustworthiness of generated responses. Experiments show that our claim-level grounding approach alleviates the limitations of conventional grounding methods in aligning generated statements with contextual sentence-level evidence, resulting in substantial improvements in overall grounding quality and user verification efficiency. The code and data are available at https://github.com/chubohao/eTracer.