[
https://issues.apache.org/jira/browse/LUCENE-4845?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13607539#comment-13607539
]
Michael McCandless commented on LUCENE-4845:
--------------------------------------------
bq. I think its because your FreeDB has a lot more words than my place names?
I think so. Song titles are longer than place names :)
bq. But really there must be a infixing limit for relevance reasons alone.
I think the app can decide this.
bq. Why is it so bad, but the edge-ngrams limit ok?
I don't think either limit is OK! In the ideal world we wouldn't require such
limits due to performance/RAM issues.
But no suggester is perfect, this is why we offer multiple options. These two
approaches have different tradeoffs...
> Add AnalyzingInfixSuggester
> ---------------------------
>
> Key: LUCENE-4845
> URL: https://issues.apache.org/jira/browse/LUCENE-4845
> Project: Lucene - Core
> Issue Type: Improvement
> Components: modules/spellchecker
> Reporter: Michael McCandless
> Assignee: Michael McCandless
> Fix For: 5.0, 4.3
>
> Attachments: infixSuggest.png, LUCENE-4845.patch, LUCENE-4845.patch,
> LUCENE-4845.patch
>
>
> Our current suggester impls do prefix matching of the incoming text
> against all compiled suggestions, but in some cases it's useful to
> allow infix matching. E.g, Netflix does infix suggestions in their
> search box.
> I did a straightforward impl, just using a normal Lucene index, and
> using PostingsHighlighter to highlight matching tokens in the
> suggestions.
> I think this likely only works well when your suggestions have a
> strong prior ranking (weight input to build), eg Netflix knows
> the popularity of movies.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]