How to combine default BM25 score of Elasticsearch and Dense Vectors similarity

Elastic Stack Elasticsearch
1

i found a issue about how to combine the both default bm25 and vectors, but i try it and found the result was poor, is there any official document about how to combine default BM25 score of Elasticsearch and Dense Vectors similarity ? including how to set the important weight. Maybe more examples will help us to understand this powerful tools.

2

There is no any official recipe how to set up these weights. You can frame this problem as as ML parameter optimization problem for your particular dataset.

I can refer you to our blog how this parameter optimization problem can be approached: Improving search relevance with data-driven query optimization

3

awesome! thank you very much

4

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.