Find and link similar entities in a knowledge graph using Amazon Neptune, Part 1: Full-text search

This article discusses using Amazon Neptune, a graph database service, for finding and linking similar entities in a knowledge graph. It covers two approaches: lexical search using full-text search, and semantic search using vector similarity (covered in Part 2).

Specifically, the article covers:

• Why similarity search is useful for knowledge graphs
• How to use Neptune's full-text search with Amazon OpenSearch Service for lexical similarity (fuzzy text matching)
• Example queries to find and link similar patient entities based on names, encounters, payers, etc.
• A convention of adding a "matches" edge between similar entities to materialize the link
• Using this approach for other entity types like publications
• Setting up and running the provided sample notebook