[Feature]: Can it support multi-vector search with weights like ES?

[weimingdiit]

Is there an existing issue for this?

• [X] I have searched the existing issues

Is your feature request related to a problem? Please describe.

In ES, we can use the "boost" parameter to indicate the weight of the input object, thus affecting our search results. like :

body={ "knn": [ { "field": "data_target_agent_feature", "query_vector": [i for i in range(256)], "k": 10, "num_candidates": 100, "boost": 0.4 }, { "field": "data_target_agent_feature", "query_vector": [i for i in range(256)], "k": 10, "num_candidates": 100, "boost": 0.6 }], "size": 50, "_source": ["file_path", "sample_index"], }

Describe the solution you'd like.

Perhaps we can use this weight to recalculate the distance according to the weight of the search results of multiple vectors, and then sort the output

Describe an alternate solution.

No response

Anything else? (Additional Context)

No response

[xiaofan-luan]

@weimingdiit The community is actually working on it. @soothing-rain @czs007

[xiaofan-luan]

let's discuss somewhere next week