[
https://issues.apache.org/jira/browse/MAHOUT-389?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12863861#action_12863861
]
Sean Owen commented on MAHOUT-389:
----------------------------------
The issue with that is you are implicitly assuming that missing ratings are 0,
and that's not a good assumption. For example, if your ratings are on a scale
of 1 to 5, you are assuming that all unknown items are utterly hated.
It is not ignoring zeroes; it is ignoring missing ratings.
I think the distributed computation should act the same way for similar reasons.
> UncenteredCosineSimilarity
> ---------------------------
>
> Key: MAHOUT-389
> URL: https://issues.apache.org/jira/browse/MAHOUT-389
> Project: Mahout
> Issue Type: Bug
> Components: Collaborative Filtering
> Reporter: Sebastian Schelter
> Attachments: MAHOUT-389.patch
>
>
> org.apache.mahout.cf.taste.impl.similarity.UncenteredCosineSimilarity only
> computes the cosine distance between those components of the vectors where
> both vectors have a value greater zero.
> This is inconsistent with the definition of the cosine (correct me if I'm
> wrong) and is inconsistent with the distributed cosine similarity computation.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
