Vector Search with the Go Client

Tip	The source code for this tutorial is available on GitHub.

This tutorial demonstrates how to perform kNN vector search and hybrid queries with the Go Hot Rod client.

Prerequisites

A running Infinispan server.
Go 1.25 or later.

Running the Example

go run ./vector-search

Code Walkthrough

Vector Schema

The Beer schema includes a @Vector field with 3 dimensions using cosine similarity.

/* @Indexed */
message Beer {
  /* @Keyword(projectable = true, sortable = true) */
  string name = 1;
  /* @Vector(dimension = 3, similarity = COSINE) */
  repeated float descriptionEmbedding = 7;
}

kNN Vector Search

Use the <→ operator to find the k nearest neighbors to a query vector.

cache.Query(ctx, "from quickstart.Beer b where b.descriptionEmbedding <-> [:v]~:k",
    hotrod.WithQueryParam("v", []float32{0.9, 0.1, 0.1}),
    hotrod.WithQueryParam("k", int32(3)),
)

Score Projection

Use score(b) to include the similarity score in projections.

cache.Query(ctx, "select b.name, b.style, score(b) from quickstart.Beer b where b.descriptionEmbedding <-> [:v]~:k",
    hotrod.WithQueryParam("v", []float32{0.05, 0.9, 0.1}),
    hotrod.WithQueryParam("k", int32(3)),
)

Hybrid Queries

Combine vector search with metadata filters using the filtering clause.

cache.Query(ctx,
    "select score(b), b.name, b.style, b.abv from quickstart.Beer b "+
    "where b.descriptionEmbedding <-> [:v]~:k "+
    "filtering (b.style = 'Lager' and b.abv < 5.0)",
    hotrod.WithQueryParam("v", []float32{0.05, 0.95, 0.05}),
    hotrod.WithQueryParam("k", int32(3)),
)

Combine vector search with full-text filters.

cache.Query(ctx,
    "select score(b), b.name from quickstart.Beer b "+
    "where b.descriptionEmbedding <-> [:v]~:k "+
    "filtering b.description : 'citrus'",
    hotrod.WithQueryParam("v", []float32{0.1, 0.1, 0.95}),
    hotrod.WithQueryParam("k", int32(5)),
)

Expected Output

The example runs 8 query demonstrations including full-text search, keyword and range filters, projections with sorting, kNN vector search, score projection, and three hybrid query variants.