Entity-Document Linking
Association of named entities extracted from unstructured documents with corresponding entries in a knowledge base or graph, enabling semantic context for Retrieval-Augmented Generation (RAG) systems.
Core process:
- Entity extraction via Named Entity Recognition
- Normalization to standardized knowledge base entries
- Context-aware linking to document semantics
Key applications:
- Enhancing rag with graph-based semantic relationships
- Enabling dynamic knowledge graph updates from document collections
- Supporting cross-document entity disambiguation
Implementation tools:
- Cocoindex: Framework for building real-time knowledge graphs from documents using large-language-models (LLMs) and Neo4j, demonstrated in this tutorial
- Neo4j: Graph database for storing entity relationships and document context
2026 04 14 Cocoindex channel and knowledge Graphs for LLM RAG