Vector Search API

The Vector Search API enables powerful semantic search capabilities by converting text into high-dimensional vectors and performing similarity comparisons using pgvector.

Base URL

https://api.koveh.com/vector-search/

Endpoints

1. Index Data

POST /index

Indexes data from a source database table into the vector store.

Fields:

• source_db_url: URI of the database to read from.
• vector_db_url: URI of the database with pgvector enabled.
• table_name: Table to index.
• text_columns: List of columns to combine for embedding.
• limit: Max records to index.

2. Perform Search

POST /search

Searches indexed data using semantic similarity.

Request Schema:

{
  "query": "Looking for a React developer job",
  "top_k": 5,
  "metric": "Cosine",
  "source_db_url": "...",
  "vector_db_url": "...",
  "res_table": "jobs",
  "res_id_col": "id",
  "res_view_cols": ["title", "company", "description"]
}

Metrics Supported:

• Cosine: Cosine similarity (Distance: 1 - Cosine).
• L2: Euclidean distance.
• InnerProduct: Negative inner product.

Usage Note

This API is used internally by our scrapers and search interfaces to provide relevant results based on meaning rather than just keywords.