lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael McCandless <luc...@mikemccandless.com>
Subject Re: indexing analyzed and not_analyzed values in same field
Date Fri, 18 Nov 2016 14:05:18 GMT
So when a query arrives, you know the query is only allowed to match
either module:1 (analyzed terms) or module:2 (not analyzed) but never
both?  If so, you should be fine.

Though relevance will be sort of wonky, in case that matters, because
you are polluting the unique term space; you would get different
relevance results if you used two separate fields instead.

Mike McCandless

http://blog.mikemccandless.com

On Fri, Nov 18, 2016 at 8:50 AM, Kumaran Ramasubramanian
<kums.134@gmail.com> wrote:
>
>
> Yes. But am going to provide search for my single module alone, so query
> will match only one type of document in any case.
>
> Here is how i use, append module:1 or module:2 in all queries i make. so
> documents that matched module:1 will have only analyzed terms and module:2
> will have only not_analyzed terms.
>
>
>
>
> --
> Kumaran R
>
>
>
> On Fri, Nov 18, 2016 at 7:04 PM, Michael McCandless
> <lucene@mikemccandless.com> wrote:
>>
>> You can do this, Lucene will let you, but  it's typically a bad idea
>> for search relevance because some documents will return only if you
>> search for precisely the same whole token, others if you search for an
>> analyzed token, giving the user a broken experience.
>>
>> Mike McCandless
>>
>> http://blog.mikemccandless.com
>>
>>
>> On Fri, Nov 18, 2016 at 6:00 AM, Kumaran Ramasubramanian
>> <kums.134@gmail.com> wrote:
>> > Hi All,
>> >
>> >    Can anyone say,  is it advisable to have index with both analyzed and
>> > not_analyzed values in one field?
>> >
>> > Use case: i have custom fields in my product which can be configured
>> > differently ( ANALYZED and NOT_ANALYZED ) in different modules
>> >
>> > --
>> > Kumaran R
>> >
>> >
>> >
>> >
>> >
>> > On Wed, Oct 26, 2016 at 12:07 AM, Kumaran Ramasubramanian <
>> > kums.134@gmail.com> wrote:
>> >
>> >>
>> >>
>> >> Hi All,
>> >>
>> >> i have indexed 4 documents in an index where BANKNAME field is analyzed
>> >> in two documents and it is not_analyzed in another two documents. i
>> >> have
>> >> mentioned search cases below where i am able to search using both
>> >> analyzed
>> >> ( using classic analyzer ) and not_analyzed ( using keyword analyzer )
>> >> terms. But, is it right to have index with both analyzed and
>> >> not_analyzed
>> >> values in a field?
>> >>
>> >>
>> >>
>> >>
>> >> output:
>> >>
>> >>
>> >> BANKNAME field of these two documents is analyzed
>> >>
>> >> using classic analyzer
>> >>  query : BANKNAME:"swiss bank"
>> >> total hits:2
>> >>
>> >> DocId:0  DocScore:1.6096026
>> >> [stored,indexed,tokenized<BANKNAME:swiss  bank>,
>> >> stored,indexed,tokenized<PLACENAME:swissland>,
>> >>
>> >> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<company:goog>]
>> >>
>> >> DocId:2  DocScore:1.6096026
>> >> [stored,indexed,tokenized<BANKNAME:swiss  bank>,
>> >> stored,indexed,tokenized<PLACENAME:swissland>,
>> >>
>> >> stored,indexed,tokenized,omitNorms,indexOptions=DOCS_ONLY<company:goog>]
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> BANKNAME field of these two documents is not analyzed
>> >>
>> >> using keyword analyzer
>> >> rrsk query : BANKNAME:swiss bank
>> >> total hits:2
>> >>
>> >> DocId:1  DocScore:1.287682
>> >> [stored,indexed,tokenized<BANKNAME:swiss bank>,
>> >> stored,indexed,tokenized<PLACENAME:swiss>,
>> >> stored,indexed,tokenized,omitN
>> >> orms,indexOptions=DOCS_ONLY<company:goog>]
>> >>
>> >> DocId:3  DocScore:1.287682
>> >> [stored,indexed,tokenized<BANKNAME:swiss bank>,
>> >> stored,indexed,tokenized<PLACENAME:swiss>,
>> >> stored,indexed,tokenized,omitN
>> >> orms,indexOptions=DOCS_ONLY<company:goog>]
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> --
>> >> Kumaran R
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message