FunctionScore

A FunctionScore instance is a list of Function instances used as rerankers.

io.milvus.v2.service.vector.request.FunctionScore

Constructor

This constructor initializes a new FunctionScore instance that comprises one or more rankers.

FunctionScore.builder()
    .functions(List<CreateCollectionReq.Function> functions)
    .params(Map<String, String> params)
    .build()

BUILDER METHODS:

• functions(List<CreateCollectionReq.Function> functions)

  A list of Function instances.

• params(Map<String, String> params)

  Extra parameters for how the specified functions work together. For a Boost ranker, you can set the following parameters:

  • boost_mode (String)

    Specifies how the specified weights influence the scores of any matching entities. Possible values are:

    • Multiple

      Indicates that the weighted value is equal to the original score of a matching entity multiplied by the specified weight.

      This is the default value.

    • Sum

      Indicates that the weighted value is equal to the sum of the original score of a matching entity and the specified weight

  • function_mode (String)

    Specifies how the weighted values from various Boost Rankers are processed. Possible values are:

    • Multiplify

      Indicates that the final score of a matching entity is equal to the product of the weighted values from all Boost Rankers.

      This is the default value.

    • Sum

      Indicates that the final score of a matching entity is equal to the sum of the weighted values from all Boost Rankers.

RETURN TYPE:

FunctionScore

RETURNS:

A FunctionScore instance.

Example

import io.milvus.common.clientenum.FunctionType;
import io.milvus.v2.service.collection.request.CreateCollectionReq;
import io.milvus.v2.service.vector.request.SearchReq;
import io.milvus.v2.service.vector.response.SearchResp;
import io.milvus.v2.service.vector.request.data.EmbeddedText;

CreateCollectionReq.Function ranker = CreateCollectionReq.Function.builder()
                 .functionType(FunctionType.RERANK)
                 .name("boost")
                 .param("reranker", "boost")
                 .param("filter", "doctype == \"abstract\"")
                 .param("weight", "0.5")
                 .param("random_score", "{\"seed\": 126, \"field\": \"id\"}")
                 .build();
                 
SearchResp searchReq = client.search(SearchReq.builder()
        .collectionName("my_collection")
        .data(Collections.singletonList(new FloatVec(new float[]{-0.619954f, 0.447943f, -0.174938f, -0.424803f, -0.864845f})))
        .annsField("vector")
        .outputFields(Collections.singletonList("doctype"))
        .functionScore(FunctionScore.builder()
                .addFunction(ranker)
                .build())
        .build());
SearchResp searchResp = client.search(searchReq);