In my current setup, I use a stemmer to analyze a field called summary. The
user is able to search over the summary field using full lucene syntax,
which we pass directly to elasticsearch through a query_string query,
inside a filter. We always sort results based on their date field, so we
use a constant score.
Unfortunately, our use of the stemmer has problems when users search for
phrases in quotations. For instance, a search for "Crowd Valley" returns
hits for the phrase "crowded valley". Ideally, I'd like to be able to
ignore the stemmer for the portion of a query in a quoted phrase, but use
the stemmer for all parts of the query outside of quotes.
Is such a thing in possible for ElasticSearch? If not, does anyone have any
best practices for getting close to this behavior as possible?
Our current best idea is to use a multi_field with two analyzers - one with
a stemmer, one without a stemmer. This would allow us to search either one
or the other, depending on if the user has a quotation mark present in
their query. However this would not be ideal for mixed queries such as:
"Crowd Valley" or running clubs [because the running clubs part would not
CTO & Co-founder, Wiser
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 firstname.lastname@example.org.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/88e6b645-73e5-4bca-9b99-082fcdc2c655%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.