Neo4j Vector Search

Neo4j Vector Search enhances the Neo4j graph database by providing vector search capabilities through dedicated indexes. These vector indexes enable similarity searches and complex analytical queries by representing nodes or properties as vectors in a multidimensional space. Originally released as a public beta in Neo4j 5.11, node vector indexes reached general availability in Neo4j 5.13.

Features

• Similarity Search: Facilitates similarity searches based on vector embeddings.
• Complex Analytical Queries: Supports complex analytical queries using vector representations.
• Powered by Apache Lucene: Leverages the Apache Lucene indexing and search library for its core functionality.
• Querying Large Datasets: Allows for efficient querying of vector embeddings within large datasets.
• Flexible Embedding Storage: Stores vector embeddings as LIST<INTEGER | FLOAT> properties on nodes or relationships.
• Broad Embedding Generator Compatibility: Compatible with embeddings generated by proprietary tools like Vertex AI and OpenAI, as well as open-source libraries such as sentence-transformers.
• Supports Various Dimensions: Accommodates vector embeddings with various dimensions, including 256, 768, 1536, and 3072.