Weaviate

Weaviate stores objects in collections with vector embeddings. Search does semantic search; fetch gets by id.

Data Model

• Collections — like tables.
• Objects — properties + vector.
• Search — vector/semantic query. Hybrid search for keyword + vector.
• Schema — cross-references.
• Batch — import, update, delete.
• Quality — duplicates, missing vectors, property validation.

Gotchas

• Resource id: weaviate:object:collection:id. Get from search/list results for fetch.
• Vector generation: May be auto by Weaviate. Check collection config.
• Cross-references: Check for reference integrity.