[
https://issues.apache.org/jira/browse/SOLR-1979?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12966964#action_12966964
]
Jan Høydahl commented on SOLR-1979:
-----------------------------------
Simply allowing to set the threshold for isReasonablyCertain() is probably not
enough to get a robust detection. This is because the distance measure is very
sensitive to the length of the profiles in use. Thus, it is a bit dangerous to
expose getDistance() as in TIKA-568, cause that distance measure is kind of an
internal value, not very normalized and is bound to change in future versions
of TIKA.
See TIKA-369 and TIKA-496.
I think the right way to go is solving these two issues first. By fixing so
that getDisance() is not biased towards profile length, we can make a new
isReasonablyCertain() implementation taking into account the relative distance
between first and second candidate languages...
> Create LanguageIdentifierUpdateProcessor
> ----------------------------------------
>
> Key: SOLR-1979
> URL: https://issues.apache.org/jira/browse/SOLR-1979
> Project: Solr
> Issue Type: New Feature
> Components: update
> Reporter: Jan Høydahl
> Assignee: Grant Ingersoll
> Priority: Minor
> Attachments: SOLR-1979.patch
>
>
> We need the ability to detect language of some random text in order to act
> upon it, such as indexing the content into language aware fields. Another
> usecase is to be able to filter/facet on language on random unstructured
> content.
> To do this, we wrap the Tika LanguageIdentifier in an UpdateProcessor. The
> processor is configurable like this:
> {code:xml}
> <processor
> class="org.apache.solr.update.processor.LanguageIdentifierUpdateProcessorFactory">
> <str name="inputFields">name,subject</str>
> <str name="outputField">language_s</str>
> <str name="idField">id</str>
> <str name="fallback">en</str>
> </processor>
> {code}
> It will then read the text from inputFields name and subject, perform
> language identification and output the ISO code for the detected language in
> the outputField. If no language was detected, fallback language is used.
--
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: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
