Pradeep, First, some background on fuzzy term expansions:
1) A query for foobar~2 is really a query for (foobar OR foobar~1 OR foobar~2) 2) Fuzzy term expansion will only take the first 50 terms found in the index and drop the rest. For implementation notes, see this comment - https://github.com/apache/lucene-solr/blob/master/lucene/core/src/java/org/apache/lucene/search/FuzzyTermsEnum.java#L229-L232 So in your first search, your available terms in the title_txt_en field are few enough that "probl~2" does match "problem" In your second search, with the copy field, there are likely many more terms in all_text_txt_enus. Here, the edit distance 1 terms crowd out the edit distance 2 terms and they never match. You can imagine that the term expands into… "probl OR prob OR probe OR prob1 OR…" I don't see a way to specify the number of expansions from a Solr query, maybe somebody else on the list would know. But at the end of the day, like Wunder said, you might want a prefix query based on what you're describing. Mike On Mon, Apr 13, 2020 at 6:01 PM Deepu <kpkumar1...@gmail.com> wrote: > Corrected Typo mistake. > > Hi Team, > > We have 8 text fields (*_txt_en) in schema and one multi valued text field > which is copy field of other text fields, like below. > > tittle_txt_en, configuration_summary_txt_en, all_text_txt_ens (multi value > field) > > Observed one issue with Fuzzy match, same term with distance of two(~2) is > working on individual fields but not returning any results from multi > valued field. > > Term we used is "probl" and document has "problem" term in two text fields, > so all_text field has two occurrences of 'problem" terms. > > > > title_txt_en:probl~2. (given results) > > all_text_txt_ens:probl~2 (no results) > > > > is there any other factors involved in distance calculation other > than Damerau-Levenshtein Distance algoritham? > > what might be the reason same input with same distance worked with one > field and failed with other field in same collection? > > is there a way we can get actual distance solr calculated w.r.t specific > document and specific field ? > > > > Thanks in advance !! > > > Thanks, > > Pradeep > > On Mon, Apr 13, 2020 at 2:35 PM Deepu <kpkumar1...@gmail.com> wrote: > > > Hi Team, > > > > We have 8 text fields (*_txt_en) in schema and one multi valued text > field > > which is copy field of other text fields, like below. > > > > tittle_txt_en, configuration_summary_txt_en, all_text_txt_ens (multi > value > > field) > > > > Observed one issue with Fuzzy match, same term with distance of two(~2) > is > > working on individual fields but not returning any results from multi > > valued field. > > > > Term we used is "prob" and document has "problem" term in two text > fields, > > so all_text field has two occurrences of 'problem" terms. > > > > > > > > title_txt_en:prob~2. (given results) > > > > all_text_txt_ens:prob~2 (no results) > > > > > > > > is there any other factors involved in distance calculation other > > than Damerau-Levenshtein Distance algoritham? > > > > what might be the reason same input with same distance worked with one > > field and failed with other field in same collection? > > > > is there a way we can get actual distance solr calculated w.r.t specific > > document and specific field ? > > > > > > > > Thanks in advance !! > > > > > > Thanks, > > > > Pradeep > > >
