KnnVectorQuery performs an approximate nearest neighbor search to find the top-k most similar vectors in a dataset.
Data structure
message KnnVectorQuery {
optional string field_name = 1;
optional int32 top_k = 2;
repeated float float32_query_vector = 4;
optional Query filter = 5;
optional float weight = 6;
optional float min_score = 7;
}|
Parameter |
Type |
Required |
Description |
|
field_name |
string |
Yes |
The name of the vector field to search. |
|
top_k |
int32 |
Yes |
The number of nearest neighbors to return. For the maximum value, see Search Index limits. |
|
float32_query_vector |
float |
Yes |
The query vector used for the similarity search. The dimension of this vector must match the dimension of the target vector field. |
|
filter |
No |
A query filter that combines any non-vector query conditions to narrow results before the vector search. |
|
|
weight |
float |
No |
The weight assigned to this query in a composite query. A higher value increases the influence of this query on the final relevance score. |
|
min_score |
float |
No |
The minimum score threshold. Only rows with a score greater than this value are returned. Valid values: greater than or equal to 0. Default value: 0. |