word2vec.js

node.js version of https://code.google.com/p/word2vec

Install

• npm install word2vec.js

Usage

Training

• training data
  • http://mattmahoney.net/dc/text8.zip -O text8.gz
• example

  var word2vec = require('word2vec.js');
  
  word2vec.trainer({
      train: './data/text8',
      output: 'vector.txt',
      on: function (log) {
          process.stdout.write(log);
      },
      done: function () {
          console.log('finish');
      },
      error: function (err) {
          console.log(err);
      }
  });

Analysis

• use word2vec.analyzer
  • analyzer.findVec(string)
  • analyzer.operator
    • vector1+vector2: analyzer.operator.sum(vector1, vector2)
    • vector1-vector2: analyzer.operator.sub(vector1, vector2)
    • distance: analyzer.operator.dist(vector1, vector2)
  • analyzer.findCousin(vector, size)
    • output like this

      [ { key: 'korea', dist: '0.771689' },
        { key: 'beijing', dist: '0.576056' },
        { key: 'seoul', dist: '0.564867' },
        { key: 'korean', dist: '0.562586' },
        { key: 'chungcheong', dist: '0.529448' },
        { key: 'dae', dist: '0.502329' },
        { key: 'vietnamese', dist: '0.497823' },
        { key: 'berlin', dist: '0.487074' },
        { key: 'chung', dist: '0.484467' },
        { key: 'gyeonggi', dist: '0.482213' } ]
        

• example

  var word2vec = require('word2vec.js');
  var analyzer = word2vec.analyzer('./vector.txt');
  
  // Find Word's Vector
  var king = analyzer.findVec('berlin');
  var man = analyzer.findVec('germany');
  var woman = analyzer.findVec('korea');
  
  // Vector Operation
  var op = analyzer.operator.sub(king, man);
  op = analyzer.operator.sum(op, woman);
  
  // Find Cousins
  var cousins = analyzer.findCousin(op, 10);
  console.log(cousins);