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

Posted to issues@spark.apache.org by "Benjamin Radford (JIRA)" <ji...@apache.org> on 2017/03/10 04:59:37 UTC

[jira] [Created] (SPARK-19892) Implement findAnalogies method for Word2VecModel

Benjamin Radford created SPARK-19892:
----------------------------------------

             Summary: Implement findAnalogies method for Word2VecModel 
                 Key: SPARK-19892
                 URL: https://issues.apache.org/jira/browse/SPARK-19892
             Project: Spark
          Issue Type: New Feature
          Components: MLlib
    Affects Versions: 2.1.0
            Reporter: Benjamin Radford
            Priority: Minor


Word2VecModel is missing a method that allows for performing analogy-like queries on word vectors (e.g. King + Woman - Man = Queen). This is a functionality common to other word2vec implementations (see gensim) and is a major component of word2vec's appeal as cited in seminal works on the model (https://code.google.com/archive/p/word2vec/). 

An implementation of this method, findAnalogies, should accept three arguments:
* positive - Array[String] of similar words
* negative - Array[String] of dissimilar words
* num - Int number of synonyms or nearest neighbors to calculated vector



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

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