Ask Questions and Find Answers

I am trying to build semantic search on top of liferay search. Do you have any ideas how to implement it in effective way ?

That's quite an innocent little line for a potentially open-ended implementation. I'd suggest to rather break it down into individual chunks - start with the underlying business problem: What are the search terms that you expect to get an improved answer for when your implementation is done? What is the current answer? What is your expectation?
And "effective" doesn't really give any additional hint: Effective for CPU? I/O? Maintenance? Initial Implementation effort?

Sorry for too short explanation of my problem which i need to resolve. 
I want to achieve effect that search will get simillar results with similar phrases to this one which user type in searchbar,
so for example if someone type "sport" in search bar then in result i should get results with articles 
containing words like "tennis", "volleyball", "football" etc.  
Currently default search mechanism returns result with articles containg word "sport" or "sporty", it can find words without suffix (-ing, -able, -ify, -y etc. )
By effective i mean initial implementation effort. I found some solution based on automatical tags generation from article content and adding simmilar tags from thesaurii dictionary but it requires a lot of time for implementation. 
Maybe liferay has some built in mechanism for achieving this effect which i didn't find.

Hello Daniel,

Even though this is an old thread, I wanted to highlight that the semantic search capabilities debuted as a beta in Liferay DXP 7.4 U70. You can learn more about it at https://learn.liferay.com/w/dxp/using-search/liferay-enterprise-search/search-experiences/semantic-search.

This recording​of our session delivered on /dev/24 last year provides a good overview of the concepts of semantic search (including a quick demo).