Vector Search

Semantic search over your knowledge graph

Vector Search allows you to find information in your world based on its meaning, rather than exact keyword matches.

How it Works

1. Ingestion: When you insert triples into your world, the API automatically generates embeddings for the subject, predicate, and object.
2. Indexing: These embeddings are stored in a vector index.
3. Querying: When you perform a search, your query is embedded, and the API finds the semantically closest triples in the index.

Use Cases

• RAG (Retrieval Augmented Generation): Retrieve relevant context for an LLM based on a user’s prompt.
• Fuzzy Finding: Find entities even if the user doesn’t know their exact URI or name.