Hippo supports creating and using index on dense and sparse vectors.
curl -u shiva:shiva -XPUT 'localhost:8902/hippo/v1/{table}/_bulk?pretty' -H 'Content-Type: application/json' -d'{
"fields_data": [
{
"field_name": "book_id",
"field": [1,2]
},
{
"field_name": "book_intro",
"field": [ [1,1,1,1],[2,1,2,2]] // 4维稠密向量
},
{
"field_name": "book_intro_sparse",
"field": [ "10:1, 200:1, 10002:1, 200000:1", "23:1, 204:1, 30000:2, 40000:1"] // 稀疏向量
}
],
"num_rows": 2,
"op_type": "insert"
}';
After creating and activating the index, users can use the same command from sparse floating-point vector to perform vector search.