If we have diacritic character in match query with fuzziness:AUTO, the terms are not expanded as expected.
This is my query and i have a standard and asciifolding analyzer on 'my_field_name' field:
{
"match": {
"my_field_name": {
"query": "crépeault",
"operator": "OR",
"fuzziness": "AUTO",
"prefix_length": 0,
"max_expansions": 50,
"fuzzy_transpositions": false,
"lenient": false,
"zero_terms_query": "NONE",
"auto_generate_synonyms_phrase_query": false,
"boost": 1.0
}
}
}
CASE 1
When i do a _validate/query?explain
I can see the generated query is:
Synonym(my_field_name:crepeault my_field_name:crépeault)
CASE 2
I change my query value to "crepeault" (Notice the diacritic is replaced by normal 'e')
I can see the generated query is:
BooleanQuery ((my_field_name:crepault)^0.875 my_field_name:crepeault (my_field_name:crepault)^0.875 (my_field_name:crepeau)^0.71428573 my_field_name:crepeault (my_field_name:crépault)^0.75 (my_field_name:crépeault)^0.8888889 (my_field_name:preseault)^0.7777778)
MY QUERY
Why does the expansion of fuzzy terms generate a SynonymQuery if a diacritic is present & why does it expand the one without diacritic into proper fuzzy terms ??
ES Version: 6.4.1