[ 
https://issues.apache.org/jira/browse/LUCENE-3370?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13163307#comment-13163307
 ] 

Trejkaz commented on LUCENE-3370:
---------------------------------

Well, I ran with a modified version of SpanNotQuery for some time and nobody 
noticed any issues with it, but I just found the one thing which SpanNotQuery 
does differently from SpanNearQuery which makes it unsuitable for this task.

With a SpanNearQuery, if you have "cat" in the document only once, and you 
search for span-near("cat","cat"), you will get no hits.  It doesn't regard 
terms as being "near" themselves.

However with a SpanNotQuery, if you have "cat" in the document only once, and 
you search for span-not("cat","cat"), you *also* get no hits, because you have 
subtracted all the spans you got in the first round.

Since SpanNotNearQuery works like an expanded SpanNotQuery, it inherits this 
behaviour.  Thus, SpanNearQuery and SpanNotNearQuery end up in a situation 
where, quite confusingly to someone who doesn't know how they work, the results 
when added together for some reason do not give the full set of spans you would 
have had before applying the additional query.

                
> Support for a "SpanNotNearQuery"
> --------------------------------
>
>                 Key: LUCENE-3370
>                 URL: https://issues.apache.org/jira/browse/LUCENE-3370
>             Project: Lucene - Java
>          Issue Type: New Feature
>          Components: core/search
>            Reporter: Trejkaz
>
> Sometimes you want to find an instance of a span which does not hit near some 
> other span query.  SpanNotQuery only excludes exact hits on the term, but 
> sometimes you want to exclude hits 1 away from the first, and other times you 
> might want the range to be wider.
> So a SpanNotNearQuery could be useful.  
> SpanNotQuery is actually very close, and adding slop+inOrder support to it is 
> probably sufficient to make a SpanNotNearQuery. :)
> There appears to be one project which has done it in this fashion, although 
> this particular code looks like it's out of date:
> http://www.koders.com/java/fid933A84488EBE1F3492B19DE01B2A4FC1D68DA258.aspx?s=ArrayQuery

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to