[jira] Commented: (SOLR-732) Collation bug

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] Commented: (SOLR-732) Collation bug

JIRA jira@apache.org

    [ https://issues.apache.org/jira/browse/SOLR-732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12891646#action_12891646 ]

James Dyer commented on SOLR-732:
---------------------------------

I do not think this is a bug.  Suggestions are ordered by score (ie. Levenstein Distance) rather than # hits.  See org.apache.lucene.search.spell.SuggestWord.compareTo() .  The "score" variable is set in org.apache.lucene.search.spell.SpellChecker.suggestSimilar.

In working with the spellchecker, if setting spellcheck.count to a high value (like 100), I've often gotten results far down the list with a lot more hits than the ones early in the list but the word is obviously a less-likely correction than the ones higher up.

Perhaps this old ticket can be closed?

> Collation bug
> -------------
>
>                 Key: SOLR-732
>                 URL: https://issues.apache.org/jira/browse/SOLR-732
>             Project: Solr
>          Issue Type: Bug
>          Components: spellchecker
>    Affects Versions: 1.3
>            Reporter: Matthew Runo
>            Priority: Minor
>
> Search term: Quicksilver... I get two suggestions...
> <lst name="suggestion">
> <int name="frequency">2</int>
> <str name="word">Quicksilver</str>
> </lst>
> <lst name="suggestion">
> <int name="frequency">220</int>
> <str name="word">Quiksilver</str>
> </lst>
> ...and it's not correctly spelled...
> <bool name="correctlySpelled">false</bool>
> ...but the collation is of the first term - not the one with the highest frequency?
> <str name="collation">Quicksilver</str>
> Other collations, for example, 'runnning' come up with more than one suggestion (cunning, running) but properly pick the 'best bet' based on frequency.

--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]