LAQuer: Localized Attribution Queries in Content-grounded Generation

ACL ARR 2025 February Submission4027 Authors

Abstract: Grounded text generation models often produce content that deviates from their source materials, requiring user verification to ensure accuracy. Existing attribution methods associate entire sentences with source documents, which can be overwhelming for users seeking to fact-check specific claims. In contrast, existing sub-sentence attribution methods may be more precise but fail to align with users' interests. In light of these limitations, we introduce Localized Attribution Queries (LAQuer), a new task that localizes specific spans of generated output to their corresponding source spans, enabling fine-grained and user-directed attribution. We introduce new methods for the LAQuer task, including prompting large language models (LLMs) and leveraging LLM representations. We then explore a modeling framework that extends existing attributed text generation methods to LAQuer. We evaluate this framework across two grounded text generation tasks: Multi-document Summarization (MDS) and Long-form Question Answering (LFQA). Our findings show that to achieve high-quality LAQuer attribution, a strong sentence-level attribution method is needed. Our contributions include: (1) proposing the LAQuer task to enhance attribution usability, (2) suggesting a modeling framework and benchmarking multiple baselines, and (3) proposing a new evaluation setting to promote future research on localized attribution in content-grounded generation.

Paper Type: Long

Research Area: Generation

Research Area Keywords: retrieval-augmented generation,text-to-text generation,multi-document summarization,user-centered design

Contribution Types: NLP engineering experiment

Languages Studied: English

Submission Number: 4027