AI RESEARCH
Align Documents to Questions: Question-Oriented Document Rewriting for Retrieval-Augmented Generation
arXiv CS.CL
•
ArXi:2604.17325v1 Announce Type: new Retrieval-Augmented Generation (RAG) enhances the factuality of Large Language Models (LLMs) by incorporating retrieved documents and/or generated context. However, LLMs often exhibit a stylistic bias when presented with mixed contexts, favoring fluent but hallucinated generated content over factually grounded yet disorganized retrieved evidence. This phenomenon reveals that the utility of retrieved information is bottlenecked by its presentation.