5  The Document Collection (db.collection)

Chapter Outline:

• 5.1. Documents & Indexing
  • The Document class: id, embedding, and metadata.
  • Indexing and Upserting: .index(doc) performs an atomic insert-or-replace.
  • Removing data: .drop(doc).
• 5.2. Vector Search (ANN)
  • Adding vectors via the Document(embedding=...) field.
  • Querying: .search(vector, top_k=N).
  • Use Case: Building a RAG system by combining text and vector search.
  • Helper: The rerank() function for hybrid search results.
• 5.3. Full-Text & Fuzzy Search
  • Full-Text Search (FTS): .match(query, on=["field.path"]).
  • Fuzzy Search: .match(query, fuzziness=2) for typo-tolerance.
• 5.4. Knowledge Graph
  • Creating relationships: .connect(source, target, label, metadata).
  • Single-hop traversal: .neighbors(doc, label).
  • Multi-hop (BFS) traversal: .walk(source, labels, depth, direction).