UNPKG

gullible

Version:

A naive Bayes text classifier.

62 lines (49 loc) 1.43 kB
Gullible ======== A naive Bayes text classifier in JavaScript. ### Installation The easiest way is to use `npm`: ```sh npm install gullible ``` Alternatively, you can clone the `git` repository: ```sh git clone https://github.com/erdiaker/gullible.git ``` ### Examples ```js // Load module var Gullible = require('gullible'); // Create a new classifier instance var c = new Gullible(); // Prepare a training set var trainSet = [ { text: 'My name is Ozymandias, king of kings: Look on my works, ye Mighty, and despair!', label: 'ozymandias' }, { text: 'A horse! A horse! My kingdom for a horse!', label: 'edward_iii' } ]; // Learn some text-class pairs c.learn(trainSet[0].text, trainSet[0].label); c.learn(trainSet[1].text, trainSet[1].label); // Classify a new text var testText = 'Hey! Wanna trade your horse for my kingdom?'; var cls = c.classify(testText); console.log(cls); // edward_iii // Estimate a score for text-class pair var scoreEd = c.estimate(testText, 'edward_iii'); var scoreOzy = c.estimate(testText, 'ozymandias'); console.log(scoreEd > scoreOzy); // true // Unlearn a previously learned sample. c.unlearn(trainSet[1].text, trainSet[1].label); var cls2 = c.classify(testText); console.log(cls2); // ozymandias // Serialize classifier as a JSON string var serialized = Gullible.toJSON(c); // Restore serialized classifier var restored = Gullible.fromJSON(serialized); ```