Hybrid Search Using Vector and Scalar Field

By specifying a Boolean expression, users can filter the scalar field of the entities during the vector search. Hippo will only return the results that match the filter condition. Scalar field indexing is an excellent way of improving the performance and accuracy rate of hybrid search.

curl -u shiva:shiva -XGET 'localhost:8902/hippo/v1/{table}/_search?pretty' -H 'Content-Type: application/json' -d'{
  "output_fields": ["book_id"],
  "search_params": {
    "anns_field": "book_intro",
    "topk": 2,
    "params": {
      "k_factor" : 100
    }
  },
  "vectors": [ [0.1,0.2], [0.3, 0.4] ],
  "dsl": "word_count >= 11000",
  "only_explain" : false
}';

Result:

{
  "num_queries" : 2,
  "top_k" : 2,
  "results" : [
    {
      "query" : 0,
      "fields_data" : [
        {
          "field_name" : "book_id",
          "field_values" : [11,12]
        }
      ],
      "scores" : [119.44999,142.24998]
    },
    {
      "query" : 1,
      "fields_data" : [
        {
          "field_name" : "book_id",
          "field_values" : [11,12]
        }
      ],
      "scores" : [114.85,137.25]
    }
  ]
}

Parameter description:

Parameters	Description	Required
table	Table name, such as "book" created in this example	Yes
database_name	Database where the table is located	No, defaults to "default" database
output_fields	Name of the field to return	Yes
anns_field	Name of the field to search on	Yes
topk	The number of records to return	Yes
params	Search parameter specific to vector index	No
embedding_index	Vector index used in current search	No, uses the first activated vector index built on the field specified in "anns_field" by default
vectors	Query vector	Yes
dsl	Scalar filter condition	Yes
only_explain	Whether to execute "explain" only	No, defaults to false

Table 36 Hybrid Search (Restful API)

Hybrid Search Using Array and Vector Fields

This chapter introduces how to conduct hybrid search using indexed array and vector fields. Regarding how to create array index, please refer to Create Array Index.

curl -u shiva:shiva -XGET 'localhost:8902/hippo/v1/book/_search?database_name={database_name}&pretty' -H'Content-Type: application/json' -d'{
  "output_fields": ["book_id"],
  "search_params": {
    "anns_field": "book_intro",
    "topk": 10
  },
  "vectors": [ [0.1,0.2], [0.3, 0.4] ],
  "dsl": "attributes in [attr5, attr8, attr9]"
}';

Result:

{
  "num_queries" : 2,
  "top_k" : 10,
  "results" : [
    {
      "query" : 0,
      "fields_data" : [
        {
          "field_name" : "book_id",
          "field_values" : [21,22,23,24,25,26,27,28,29,30],
          "scores" : [437.45,480.25,525.05,571.85,620.65,671.45,724.25,779.05,835.85,894.64996]
        }
      ]
    },
    {
      "query" : 1,
      "fields_data" : [
        {
          "field_name" : "book_id",
          "field_values" : [21,22,23,24,25,26,27,28,29,30],
          "scores" : [428.85,471.25003,515.65,562.05005,610.45,660.85004,713.25,767.65,824.05005,882.45]
        }
      ]
    }
  ]
}