[
https://issues.apache.org/jira/browse/MAHOUT-902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13162398#comment-13162398
]
Sean Owen commented on MAHOUT-902:
----------------------------------
Sebastian is this turned around? yes the non-distributed implementation already
has this behavior. I understood from your message that you proposed to make the
distributed version return the same thing.
I think it's as simple as making its similarity computation into:
return dots == 0 ? Double.NaN : dots / (normA + normB - dots);
> TanimotoCoefficientSimilarity should return Double.NaN for two items that
> have zero overlap
> -------------------------------------------------------------------------------------------
>
> Key: MAHOUT-902
> URL: https://issues.apache.org/jira/browse/MAHOUT-902
> Project: Mahout
> Issue Type: Improvement
> Components: Collaborative Filtering
> Affects Versions: 0.6
> Reporter: Sebastian Schelter
> Labels: MAHOUT_INTRO_CONTRIBUTE
>
> org.apache.mahout.cf.taste.impl.similarity.TanimotoCoefficientSimilarity
> should return Double.NaN for two items that have zero overlap. Please also
> provide a unit-test for this.
--
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
