Hi,
I have built an index for address searching. All address data goes to single field hooked up to an edge ngram analyzer (I set min: 1, max: 25)... It doesn't work bad, but it could work a lot better.
The problem that I have is this:
For indexed values:
FooXXXXXX
FooBarXXX
FooBarBaz
When I multi_match search for FooBarBaz (fuzziness: 0) the index will return all 3 results with not big enough a score difference.
I understand why it does so, but is there any way for it to boost the score of FooBarBaz document (over the ones found using part of the token) more than it does now?
That would be a case of boosting documents that are at least as long as the query token (not shorter).