Neo4j

Category: Vector Database Engines
Tags: graph-database, vector-search, rag, ai

---

Description

Neo4j is a graph database that now includes vector search capabilities, making it suitable for retrieval augmented generation (RAG) and other AI applications.

---

Features

• Graph Database: Native graph data storage and querying using Cypher.
• Vector Search Index: As of v5.13, Neo4j supports a dedicated Vector Search Index.
• Embedding Storage: Embeddings can be stored as properties on nodes (e.g., in an "embedding" property).
• Similarity Metrics: Supports both Euclidean and Cosine Similarity for vector similarity searches.
• Cypher Integration: Vector search and index management are performed via Cypher commands:
  • db.index.vector.createNodeIndex for creating vector indexes.
  • db.create.setNodeVectorProperty for storing vector data on nodes.
  • db.index.vector.queryNodes for querying similar nodes based on vector similarity.
• Flexible Schema: Can be adapted for various use cases, such as semantic search in note-taking applications.
• Top-K Search: Retrieve top-N similar items ranked by similarity score.

---

Pricing

No pricing information provided in the available content.

---

Source

Neo4j Vector Search - Ken Wagatsuma's Homepage