Machine Learning

Javascript

  • Overview of open source libraries
  • Recommend textual content

Open source libs

  • Natural
  • ConvNetJS
  • (statistics libs)
  • ???

var natural = require('natural'),
    classifier = new natural.BayesClassifier();


classifier.addDocument('my unit-tests failed.', 'software');
classifier.addDocument('tried the program, but it was buggy.', 'software');
classifier.addDocument('the drive has a 2TB capacity.', 'hardware');
classifier.addDocument('i need a new power supply.', 'hardware');

classifier.train();

classifier.save('classifier.json', function(err, classifier) {
    // the classifier is saved to the classifier.json file!
    
    natural.BayesClassifier.load('classifier.json', null, function(err, classifier) {
        console.log(classifier.classify('did you buy a new drive?'));
    });
});
"The Deep Learning Conspiracy"

Takeaway

  • Algorithm vs. Model
  • Javascript is good to run the Model

Let's write a model

  • Data gathering and cleaning
  • Algorithm choice
  • Model training
  • Recommend

Get data

Italian food

SELECT DISTINCT ?id ?label ?content WHERE {
    ?id a dbo:Food .
    ?id dbo:origin dbr:Italy .
    ?id dbo:abstract ?content .
    ?id rdfs:label ?label .

    FILTER(LANG(?label) = "en")
    FILTER(LANG(?content) = "en")
} LIMIT 10000
Run on dbPedia

Algorithm

Jaccard/Tanimoto index

Algorithm

Jaccard/Tanimoto index

function similarity( bagA, bagB ) {

    var nIntersection = _.intersection( bagA, bagB).length;
    var nUnion        = _.union( bagA, bagB ).length;

    return nIntersection / nUnion;
};

Similarity graph


var similarityGraph =  {
    "A":{
            "B": 0.1,
            "C": 0.8,
        },
    "B":{
            "A": 0.1,
            "C": 0.5,
        },
    "C":{
            "A": 0.8,
            "B": 0.5,
        }
};
Show me the code!