[
https://issues.apache.org/jira/browse/MAHOUT-1591?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14052825#comment-14052825
]
Ted Dunning commented on MAHOUT-1591:
-------------------------------------
Xiaomeng,
Could you direct your attention to the newer framework that avoids map-reduce?
The older framework is deprecated and optimization of it is less important.
Your assistance on the newer code would be a great help.
> Parallel job in recommendation
> ------------------------------
>
> Key: MAHOUT-1591
> URL: https://issues.apache.org/jira/browse/MAHOUT-1591
> Project: Mahout
> Issue Type: Improvement
> Reporter: Xiaomeng Huang
>
> Now all job in recommendation are sequential. But some jobs are not depend on
> the previous job. For example, in PreparePreferenceMatrixJob.java, the input
> of job itemIDIndex and toUserVectors are same. So they can run concurrently.
> And so as RowSimilarityJob.java.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
