You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@spark.apache.org by "Eric Li (JIRA)" <ji...@apache.org> on 2015/05/13 23:37:00 UTC

[jira] [Created] (SPARK-7618) Word2VecModel cache normalized wordVectors to speed up findSynonyms

Eric Li created SPARK-7618:
------------------------------

             Summary: Word2VecModel cache normalized wordVectors to speed up findSynonyms
                 Key: SPARK-7618
                 URL: https://issues.apache.org/jira/browse/SPARK-7618
             Project: Spark
          Issue Type: Improvement
          Components: MLlib
    Affects Versions: 1.3.1
            Reporter: Eric Li
            Priority: Minor


In current implementation, each findSynonyms call will need to do a Euclidean Normalization (cosineVec / wordVecNorms), this is expensive. Caching a copy of normalized wordVectors will speed up multiple findSynonyms call. This is how the Google's word2vec C code implemented. 

In addition, doing a lazy loading for wordVectors will be nice as well. 



--
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