Learning to Explore and Select for Coverage-Conditioned Retrieval-Augmented Generation

TL;DR

This paper introduces QTree and QPlanner to improve retrieval-augmented generation by generating structured outlines that better meet user-specified coverage conditions, enhancing response relevance and reducing redundancy.

Contribution

The paper presents a new dataset (QTree) and a trained model (QPlanner) for generating tailored outlines to guide LLM responses in coverage-conditioned retrieval tasks.

Findings

QPlanner produces higher-quality, user-aligned outlines.

Alignment techniques like DPO improve outline generation.

Structured outlines enhance RAG system performance.

Abstract

Interactions with large language models (LLMs) often yield long and detailed responses, leveraging both parametric knowledge and retrieval-augmented generation (RAG). While these responses can provide rich insights, they often include redundant or less engaging content not aligned with user interests. This issue becomes apparent when users specify particular subtopics to include or exclude -- termed coverage-conditioned ($C^2$) queries -- as LLMs often struggle to provide tailored responses. To address this challenge, we investigate the role of query outlines, sequences of subqueries designed to guide LLMs in generating responses that meet specific user requirements. To systematically create and evaluate these outlines, we introduce QTree, a dataset of 10K hierarchical sets of information-seeking subqueries that define structured boundaries for outline creation and evaluation in $C^2$ scenarios. Additionally, we develop QPlanner, a 7B language model trained to generate customized outlines within boundaries of QTree. We evaluate the effectiveness of the generated outlines through automatic and human judgements, focusing on their impact within retrieval-augmented generation (RAG) systems. Experimental results demonstrate that QPlanner, especially when trained with alignment techniques like DPO, generates higher-quality outlines that better fulfill diverse user needs.