[
https://issues.apache.org/jira/browse/SPARK-13938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15205026#comment-15205026
]
Steve Weng commented on SPARK-13938:
------------------------------------
Hey Sean, what classifies as whether or not features would be appropriate for
Spark? The word2phrase algo seems fairly useful and often provided along with
word2vec, which Spark does support. (For example, Gensim features phrase on
top of word2vec).
Cheers,
Steve
> word2phrase feature created in ML
> ---------------------------------
>
> Key: SPARK-13938
> URL: https://issues.apache.org/jira/browse/SPARK-13938
> Project: Spark
> Issue Type: New Feature
> Components: ML
> Reporter: Steve Weng
> Priority: Critical
> Original Estimate: 840h
> Remaining Estimate: 840h
>
> I implemented word2phrase (see http://arxiv.org/pdf/1310.4546.pdf) which
> transforms a sentence of words into one where certain individual consecutive
> words are concatenated by using a training model/estimator (e.g. "I went to
> New York" becomes "I went to new_york").
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
