On the Merits of LLM-Based Corpus Enrichment

2025-06-06

AI Summary

This work addresses the problem of limited document corpus coverage in retrieval systems, which constrains query matching accuracy. We propose an LLM-driven, active corpus augmentation paradigm that integrates generative AI as a *pre-retrieval* corpus optimization module—not merely a post-hoc refinement step. Our method combines controllable text generation, retrieval-relevance-guided prompt engineering, corpus re-ranking, and tight integration with RAG frameworks to augment the original corpus via rewriting or generating highly relevant documents. The core contribution is the first systematic application of large language models for *active, corpus-level enhancement*. Experiments on standard benchmarks (e.g., MS MARCO) demonstrate significant improvements in MRR and NDCG. In RAG settings, our approach reduces hallucination rates and enhances answer traceability and attribution reliability.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Generative AI (genAI) technologies -- specifically, large language models (LLMs) -- and search have evolving relations. We argue for a novel perspective: using genAI to enrich a document corpus so as to improve query-based retrieval effectiveness. The enrichment is based on modifying existing documents or generating new ones. As an empirical proof of concept, we use LLMs to generate documents relevant to a topic which are more retrievable than existing ones. In addition, we demonstrate the potential merits of using corpus enrichment for retrieval augmented generation (RAG) and answer attribution in question answering.

Problem

• Enhancing document retrieval via LLM-based corpus enrichment
• Improving retrieval effectiveness by modifying or generating documents
• Exploring corpus enrichment benefits for RAG and answer attribution

Innovation

• Using LLMs to enrich document corpus
• Modifying or generating documents for retrieval
• Enhancing RAG and answer attribution