Oracle Vector Store icon

Oracle Vector Store

Gerenciamento de vector store usando Oracle Database 23ai com suporte nativo a vetores

Actions7

Search Similarity

AI-generated

Summary

Queries an Oracle Database 23ai vector collection to find documents similar to a provided search vector based on specified distance metrics and threshold bounds.

Inputs

  • collectionName — The name of the Oracle table acting as the vector collection. Defaults to 'VECTOR_STORE'.
  • searchVector (required) — A JSON array string representing the embedding query vector (e.g. '[0.1, 0.2, 0.3]'). Must contain non-empty numbers.
  • limit — Maximum number of vector similarity results to return, clamped between 1 and 1000. Defaults to 10.
  • threshold — Minimum similarity score threshold (0.0 to 1.0) required for results to be included. Defaults to 0.7.
  • distanceMetric — The vector distance metric to evaluate: COSINE, EUCLIDEAN, or DOT. Defaults to COSINE.

Output shape

a list of items

Returns array items containing document details (id, content, metadata object, createdAt, updatedAt) alongside calculated vector distance and similarity score.

Examples

Example 1: Search for top 5 similar documents using cosine similarity

Collection Name: 'KNOWLEDGE_BASE', Search Vector: '[0.12, -0.45, 0.88]', Limit: 5, Threshold: 0.8, Distance Metric: 'COSINE'

Links

Discussion