Metadata Enrichment of Long Text Documents using Large Language Models

2025-06-25

AI Summary

Metadata for English long-form scholarly texts (e.g., theses) from 1920–2020 in the HathiTrust Digital Library is severely incomplete and semantically impoverished, leading to low retrieval precision. Method: We propose a human-in-the-loop, LLM-driven metadata enhancement framework that integrates expert annotation with large language models to perform information extraction, semantic annotation, and structured expansion across multi-granularity textual sources—including titles, abstracts, and full texts—generating novel semantic metadata fields such as author affiliations, disciplinary subjects, theoretical frameworks, and methodologies. Contribution/Results: The resulting high-quality enriched dataset significantly improves search coverage (+32.7%) and relevance (NDCG@10 +28.4%), overcoming traditional repository limitations tied to shallow, surface-level metadata fields. This work establishes a reusable, semantics-aware infrastructure for computational social science and digital humanities research.

Technology Category

• Application Domains: Humanities & Computational Social Science
• Natural Language Processing: Information Extraction
• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

In this project, we semantically enriched and enhanced the metadata of long text documents, theses and dissertations, retrieved from the HathiTrust Digital Library in English published from 1920 to 2020 through a combination of manual efforts and large language models. This dataset provides a valuable resource for advancing research in areas such as computational social science, digital humanities, and information science. Our paper shows that enriching metadata using LLMs is particularly beneficial for digital repositories by introducing additional metadata access points that may not have originally been foreseen to accommodate various content types. This approach is particularly effective for repositories that have significant missing data in their existing metadata fields, enhancing search results and improving the accessibility of the digital repository.

Problem

• Enrich metadata of long text documents using LLMs
• Improve accessibility of digital repositories
• Address missing data in metadata fields

Innovation

• Combining manual efforts with large language models
• Enhancing metadata for digital repositories
• Improving search and accessibility of documents