Elastic app search synonyms

Hi,
Can share what is the best practice to add synonyms?
i.e. car, it can mean motor, vehicle as well
So when user search car, I want vehicle and motor are included as well.
Should I add it as synonyms?
There is default limit of 256.
Although it can be adjusted, it seems slow down the performance.

Thanks

Welcome to the community @lostinroom !

You have the right use of synonyms, it's to define words that should always be associated with each other. You're also correct that too many App Search synonyms will have an undesirable impact on performance.

We've addressed this in the Stack with our new Synonyms API first launched in 8.10, however this synonyms API will not work out of the box with App Search. If you're spinning up a new Engine, I'd recommend looking at something like search applications instead, to allow you to use all of the features in Elasticsearch that aren't supported by App Search.

Hi,

Thanks for quick response.

The reason I use app search because it has UI to set relevance tuning and curation to to adjust the result. And also analytics to see top and recent queries.

Is there any plan to fix the performance of app search in the future?

If I understand correctly, if I use Synonyms API that you suggested, I cannot use app search. I wonder if feature like relevance tuning, curation and analytics are available without app search?

Thanks

You're correct, if you want to use App Search it won't work with the new Synonyms API and you will have to use the Synoynms API that is baked into App Search.

We have published a comparison table between App Search and native Elasticsearch as well as a [Python notebook]( an example Python notebook) that shows how you can migrate existing App Search engines to Elasticsearch.

We do have features such as query rules and behavioral analytics that aim to meet some of the functionality that App Search offers. While Elasticsearch doesn't have built in text search relevance adjustment in the manner that App Search does, it provides much more relevant out of the box search for use cases such as semantic search.