# Why relevance score becomes so high in ES 5.2?

**URL:** <https://discuss.elastic.co/t/why-relevance-score-becomes-so-high-in-es-5-2/76887>\
**Category:** Elasticsearch\
**Created:** [March 1, 2017, 12:25am UTC](https://discuss.elastic.co/t/why-relevance-score-becomes-so-high-in-es-5-2/76887 "2017-03-01T00:25:29Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Youxu](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/youxu/32/117375_2.png) [@Youxu](https://discuss.elastic.co/u/Youxu)\
**Post date:** [March 1, 2017, 12:25am UTC](https://discuss.elastic.co/t/why-relevance-score-becomes-so-high-in-es-5-2/76887/1 "2017-03-01T00:25:29Z")

</div>

In our old ES 1.x & 2.x cluster, the max relevance score is generally less than 3, but on ES 5.2, I saw very high relevance score that greater than 200.

Is it because the default similarity algorithm changed to BM 25 on ES 5.2? Is there a easy way to normalize the score to be similar to the score in ES 1.x & 2.x?

---

<div class="post-metadata">

**Author:** ![softwaredoug](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/softwaredoug/32/22681_2.png) [@softwaredoug](https://discuss.elastic.co/u/softwaredoug)\
**Post date:** [March 1, 2017, 12:51am UTC](https://discuss.elastic.co/t/why-relevance-score-becomes-so-high-in-es-5-2/76887/2 "2017-03-01T00:51:02Z")

</div>

This is because the default similarity is now BM25. The scores seem to tend  
higher. Generally it's nothing to be concerned about

More here

> **[BM25 The Next Generation of Lucene Relevance](http://opensourceconnections.com/blog/2015/10/16/bm25-the-next-generation-of-lucene-relevation/)**
>
> There’s something new cooking in how Lucene scores documents. Instead of the traditional ‘TF\*IDF’ Lucene just switched to something called BM25 in trunk. Tha...

You can change back to the classic TF\*IDF score here:  
[https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules-similarity.html](https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules-similarity.html)

But I'd ask yourself why? BM25 is generally a better general purpose  
scoring solution based on information retrieval research

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\
**Post date:** [March 29, 2017, 12:51am UTC](https://discuss.elastic.co/t/why-relevance-score-becomes-so-high-in-es-5-2/76887/3 "2017-03-29T00:51:04Z")

</div>

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