String analyzer behaves unexpectedly


(Gaurav Arora) #1

I have a custom analyzer which I use on the field name. This analyzer uses standard, lowercase, stop, trim and snowball filters. The tokenizer used is standard.

I have two documents which are:

  1. Bikanervala
  2. State Bank of Bikaner

When I search for the term, 'bikaner' , shouldn't both these records match and not just the second one? Am I using the wrong analyzer?


(sandro) #2

Just test it (it= your analyzer) with:
GET my_index/_analyze?field=name
{
"text": [...your 2 example.s..."]
}

then you see exactly what you get.


(Mark Walkom) #3

I'm not 100% sure that snowball would know what to do with Bikanervala, what language is it from?


(system) #4