0

how can we use ElasticSearch to prioritize the "correct" result groups based on the search word? ie. "italian food" = restaurants (group of data, it can be a category). did ElasticSearch by default has something like that functionality, or it must be 100% custom ?

For Search we have implemented standart ES client 

RestHighLevelClient

Maybe Anyone knows a better client to using search - currently in this situation for that type of searching?

Thank You!

Danik
  • 1
  • 2
  • Have you heard about NLP? Here is a good read on this topic: https://www.elastic.co/blog/text-classification-made-easy-with-elasticsearch – ibexit Jun 11 '19 at 21:48
  • no i don't know this NLP i found only one possibility take it together with AI and Recommendation https://qbox.io/blog/building-simple-recommender-systems-for-elasticsearch-1 maybe you tried this? Or for that solution it is not a good idea? – Danik Jun 13 '19 at 06:18
  • Just to be on the same page: theren are some categories containing some items. If users search for some items, the categories should be suggested. The 'guess' for this suggestion should be somehow educated and the 'best matching' categories are based on the given user input? Correct? – ibexit Jun 13 '19 at 15:28
  • Yes! You are correct on 100 % ! :) – Danik Jun 15 '19 at 16:28
  • There is also a simple approach in this case: Do a search for the items using the user input (italian food). The matching items needs to be aggregated on the categories. Now you have a aggregation bucket containing the categories with similar items, probably the best for the suggestion. – ibexit Jun 16 '19 at 09:04
  • And a diffreren approach using suggester: https://stackoverflow.com/questions/34903150/implementing-suggestions-xxx-in-category-using-elasticsearch – ibexit Jun 16 '19 at 09:10

0 Answers0