I'm looking into elasticsearch to replace our current engine. The most
challenging feature is the 'free text search' and finding relevant hits.
For our catalogue wide search we currently use 48 fields to search
through like producttitle, author, brand, synonyms, keywords, color,
platform, etc... I need to verify that we indeed need all those fields but
let's assume that we do
My idea is to create several indexed fields per field. So for example
producttitle gets producttitle.not_analyzed (for exact matches),
producttitle.partial (for more fuzzy matches), etc... But there are some
many options to choose from regarding filters, both on the index as on the
The set op attributes to search on can be grouped:
- productoriented attributes --> title, color
- creators of the product --> names of people
- creators of the product --> brand
- meta information --> synonyms, keywords
But i think each groups need there own set of filters but i'm looking for *advice
*on which filters to use and when. Right now i started simple with the
stemming, stopword and lowercase analyzer on index time. On index time
because in my experience doing more work upfront helps the performance when
searching. I also have a good idea which fields are more important so i can
boost on them (producttitle.not_analyzed^20)
Hope someone can give me some good pointers based on their experience.
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to email@example.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/183c9751-90a2-4b35-98f9-077dfde8f481%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.