[jira] [Commented] (SPARK-7617) Word2VecModel fVector not normalized

["Apache Spark (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-7617?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14549228#comment-14549228 ] 

Apache Spark commented on SPARK-7617:
-------------------------------------

User 'ezli' has created a pull request for this issue:
https://github.com/apache/spark/pull/6245

> Word2VecModel fVector not normalized
> ------------------------------------
>
>                 Key: SPARK-7617
>                 URL: https://issues.apache.org/jira/browse/SPARK-7617
>             Project: Spark
>          Issue Type: Improvement
>          Components: MLlib
>    Affects Versions: 1.3.1
>            Reporter: Eric Li
>            Priority: Trivial
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> fVector is not divided by norm in findSynonyms. This will not affect the order of synonyms, since it is a constant. However, the calculated cosine distances will not be comparable across different words. 
> For example, 
> findSynonyms("sf", 1) return [("oakland", 0.2)]
> findSynonyms("dc", 1) return [("arlington", 0.3)]
> We cannot say compare the distances 0.2 with 0.3, since distances are not normalized. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org