Best way to build DidYouMean feature

Hi, i've an search service based on elasticsearch.

I need to build the DidYouMean feature based on a filter and a simple requirement:

  • match against two array filter
  • correct the word for produce secure result (the word innested in a query must produce one or more result)
  • suggest from multiple fields

I've seen two way:
1 - Use phrase autocomplete with match_against feature (on standard tokenized fields)
2 - Use search with fuzzy (on standard tokenized fields)

With the first solution, I must make multiple autocomplete request in the second one (autocomplete doesn't permit multiple fields in request)
With the first solution, I need to make applicative logic to check the result.

What is the best way to perform this search? (performance & CPU load in elasticsearch)
Is correct to use the standard tokenized field for suggestion?

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