# Hybrid search without RRF

**URL:** <https://discuss.elastic.co/t/hybrid-search-without-rrf/369280>\
**Category:** Elasticsearch\
**Created:** [October 23, 2024, 11:54am UTC](https://discuss.elastic.co/t/hybrid-search-without-rrf/369280 "2024-10-23T11:54:13Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![ravitejavemula333](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/ravitejavemula333/32/138623_2.png) [@ravitejavemula333](https://discuss.elastic.co/u/ravitejavemula333)\
**Post date:** [October 23, 2024, 11:54am UTC](https://discuss.elastic.co/t/hybrid-search-without-rrf/369280/1 "2024-10-23T11:54:13Z")

</div>

Hi team, I want to do match Phrase and and also do a semantic search that means hybrid search , my current plan do not have rrf scoring mechanism, so i want to get scoring by customising it, along with i have additional filters which should filter both my match phrase results and semantic results. I am exploring in the internet but i could not find a perfect query to balance my problem. Kindly help me on this .  
example : I want to get all the docs from country germany with "backend" text to search for match\_phrase and for semantic search , it should return with me a balance scoring which i can apply threshold

---

<div class="post-metadata">

**Author:** ![Carlos\_D](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/carlos_d/32/126245_2.png) [@Carlos\_D](https://discuss.elastic.co/u/Carlos_D)\
**Post date:** [October 28, 2024, 4:09pm UTC](https://discuss.elastic.co/t/hybrid-search-without-rrf/369280/2 "2024-10-28T16:09:20Z")

</div>

Hi @ravitejavemula333 :

You can perform hybrid search without RRF by [boosting](https://www.elastic.co/guide/en/elasticsearch/reference/current/knn-search.html#_combine_approximate_knn_with_other_features) the individual queries so their scores are modified before being combined.

There is not a way of automatically handling boosting for hybrid search. You will need to experiment in order to choose appropriate values for boosting, as Elasticsearch does not provide automatic score normalization.

Hope that helps!
