Find and link similar entities in a knowledge graph using Amazon Neptune, Part 2: Vector similarity search

This article discusses how to find and link similar entities in a knowledge graph using Amazon Neptune, a managed graph database service. Specifically, it focuses on using semantic search to identify entities with similar meanings based on vector embeddings.

Specifically, the article covers:

• Semantic search using machine learning models to create vector embeddings for entities, where similar vectors represent similar meanings.
• Using the DBLP ACM dataset with publication data and generating embeddings for publication titles.
• Loading publication data and embeddings into a Neptune Analytics graph.
• Using the `topKByNode` and `topKByEmbedding` functions in Neptune Analytics to find similar publications based on embeddings.
• Comparing publication similarities along with author information to link potential duplicates.
• Using OpenSearch Service for semantic search and linking it with Neptune in a future post.
• Conclusion: Semantic search and vector embeddings can help identify and link similar entities in a knowledge graph.