[
https://issues.apache.org/jira/browse/LUCENE-3723?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Robert Muir updated LUCENE-3723:
--------------------------------
Attachment: LUCENE-3723.patch
Patch enabling the assertions, and disabling FieldMaskingSpanQuery's query
tests unless the sim for the masked field is TFIDFSimilarity.
I also changed the javadoc:
{noformat}
* <p>Note: as {@link #getField()} returns the masked field, scoring will be
- * done using the norms of the field name supplied. This may lead to unexpected
- * scoring behaviour.</p>
+ * done using the Similarity and collection statistics of the field name
supplied,
+ * but with the term statistics of the real field. This may lead to exceptions,
+ * poor performance, and unexpected scoring behaviour.</p>
*/
{noformat}
I think because this query doesn't really work except with certain scoring
models, we should at least move it to contrib if we aren't going to fix its
scoring or remove it.
> Remove FieldMaskingSpanQuery (or fix its scoring)
> -------------------------------------------------
>
> Key: LUCENE-3723
> URL: https://issues.apache.org/jira/browse/LUCENE-3723
> Project: Lucene - Java
> Issue Type: Bug
> Affects Versions: 4.0
> Reporter: Robert Muir
> Attachments: LUCENE-3723.patch, LUCENE-3723_test.patch
>
>
> In Lucene 4.0 we added new scoring mechanisms, but FieldMaskingSpanQuery is a
> serious problem:
> Because it lies about the fields of its terms, this sometimes results in
> totally bogus
> statistics, cases where a single terms totalTermFreq exceeds sumTotalTermFreq
> for the entire field (since its lying about it).
> Such lying could result in NaN/Inf/Negative scores, exceptions, divide by
> zero, and other problems,
> because the statistics are impossibly bogus.
--
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: [email protected]
For additional commands, e-mail: [email protected]
