Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Systematic Framework for Enterprise Knowledge Retrieval: Leveraging LLM-Generated Metadata to Enhance RAG Systems

Pranav Pushkar Mishra, Kranti Prakash Yeole, Ramyashree Keshavamurthy, Mokshit Bharat Surana, Fatemeh Sarayloo

TL;DR

Efficient retrieval from large, dynamic enterprise knowledge bases is essential, and current LLMs face limitations in domain-specific accuracy. The authors propose a metadata-enrichment framework that uses LLM-generated metadata to augment RAG systems, systematically evaluating three chunking strategies and multiple embedding configurations with cross-encoder ground truth. Key findings show that metadata enrichment improves retrieval precision and clustering quality, with recursive chunking plus TF-IDF weighting delivering the best overall performance, while naive chunking with prefix-fusion achieves the highest Hit Rate@10 in some setups. The work provides practical guidelines for deploying high-performance, metadata-optimized RAG architectures in enterprise knowledge management contexts.

Abstract

In enterprise settings, efficiently retrieving relevant information from large and complex knowledge bases is essential for operational productivity and informed decision-making. This research presents a systematic framework for metadata enrichment using large language models (LLMs) to enhance document retrieval in Retrieval-Augmented Generation (RAG) systems. Our approach employs a comprehensive, structured pipeline that dynamically generates meaningful metadata for document segments, substantially improving their semantic representations and retrieval accuracy. Through extensive experiments, we compare three chunking strategies-semantic, recursive, and naive-and evaluate their effectiveness when combined with advanced embedding techniques. The results demonstrate that metadata-enriched approaches consistently outperform content-only baselines, with recursive chunking paired with TF-IDF weighted embeddings yielding an 82.5% precision rate compared to 73.3% for semantic content-only approaches. The naive chunking strategy with prefix-fusion achieved the highest Hit Rate@10 of 0.925. Our evaluation employs cross-encoder reranking for ground truth generation, enabling rigorous assessment via Hit Rate and Metadata Consistency metrics. These findings confirm that metadata enrichment enhances vector clustering quality while reducing retrieval latency, making it a key optimization for RAG systems across knowledge domains. This work offers practical insights for deploying high-performance, scalable document retrieval solutions in enterprise settings, demonstrating that metadata enrichment is a powerful approach for enhancing RAG effectiveness.

A Systematic Framework for Enterprise Knowledge Retrieval: Leveraging LLM-Generated Metadata to Enhance RAG Systems

TL;DR

Efficient retrieval from large, dynamic enterprise knowledge bases is essential, and current LLMs face limitations in domain-specific accuracy. The authors propose a metadata-enrichment framework that uses LLM-generated metadata to augment RAG systems, systematically evaluating three chunking strategies and multiple embedding configurations with cross-encoder ground truth. Key findings show that metadata enrichment improves retrieval precision and clustering quality, with recursive chunking plus TF-IDF weighting delivering the best overall performance, while naive chunking with prefix-fusion achieves the highest Hit Rate@10 in some setups. The work provides practical guidelines for deploying high-performance, metadata-optimized RAG architectures in enterprise knowledge management contexts.

Abstract

In enterprise settings, efficiently retrieving relevant information from large and complex knowledge bases is essential for operational productivity and informed decision-making. This research presents a systematic framework for metadata enrichment using large language models (LLMs) to enhance document retrieval in Retrieval-Augmented Generation (RAG) systems. Our approach employs a comprehensive, structured pipeline that dynamically generates meaningful metadata for document segments, substantially improving their semantic representations and retrieval accuracy. Through extensive experiments, we compare three chunking strategies-semantic, recursive, and naive-and evaluate their effectiveness when combined with advanced embedding techniques. The results demonstrate that metadata-enriched approaches consistently outperform content-only baselines, with recursive chunking paired with TF-IDF weighted embeddings yielding an 82.5% precision rate compared to 73.3% for semantic content-only approaches. The naive chunking strategy with prefix-fusion achieved the highest Hit Rate@10 of 0.925. Our evaluation employs cross-encoder reranking for ground truth generation, enabling rigorous assessment via Hit Rate and Metadata Consistency metrics. These findings confirm that metadata enrichment enhances vector clustering quality while reducing retrieval latency, making it a key optimization for RAG systems across knowledge domains. This work offers practical insights for deploying high-performance, scalable document retrieval solutions in enterprise settings, demonstrating that metadata enrichment is a powerful approach for enhancing RAG effectiveness.

Paper Structure

This paper contains 26 sections, 3 figures, 4 tables.

Table of Contents

  1. INTRODUCTION
  2. Related Work
  3. Advanced RAG Systems and Techniques
  4. Metadata Enrichment with Large Language Models (LLMs)
  5. Advancements in Retrieval-Augmented Generation (RAG)
  6. Embedding Optimization for Semantic Search
  7. Advancements in Search System Architectures
  8. Methodology
  9. Document Processing and Chunking Strategies
  10. Naive Chunking
  11. Recursive Chunking
  12. Semantic Chunking
  13. Metadata Enrichment Pipeline
  14. Embedding Techniques
  15. Snowflake Arctic-Embed Model
  16. ...and 11 more sections

Figures (3)

  • Figure 1: The proposed system architecture for metadata enrichment and retrieval optimization .
  • Figure 2: RAG enhancement pipeline showing the progression from document chunking strategies (left), through metadata enrichment (center), to embedding generation and retrieval methods (right).
  • Figure 3: Evaluation framework for RAG systems, covering retrieval metrics and response quality to assess both retrieval effectiveness and LLM integration.