Class Feature Centroid Classifier

This is a simple machine learning algorithm for text categorization based in the Hu Guan et al. (available here) article.

$ npm i cfc-classifier

Usage

const CFC = require('cfc-classifier')

// Your dataset
const categories = ['a', 'b']
const corpus = [['category A'], ['category B']]

// Create a new classifier instance
const cfc = new CFC(categories, corpus)

// Train the classifier
cfc.train()

// Now you can classify texts
// the function below will return 'a'
cfc.classify('this text will be classified at category A')

Parsing step

In function of remove stopwords, word clustering or things like that this lib is open to insert any parsing steps that you want. In the example below I am inserting a parsing function that only remove the 'a' tokens.

const CFC = require('cfc-classifier')

const categories = ['a']
const corpus = [['a simple text, with some! interesting. things']]
const cfc = new CFC(categories, corpus)

// Add a parsing stepthis could be a remove
// stopwords function or something like that
const removeLetterA = (textTokens) => textTokens.filter(token => token.toLowerCase() !== 'a')
cfc.addParsingStep(removeLetterA)

const tokens = cfc.generateTokens(cfc.corpus)

// tokens === [ 'simple', 'text', 'with', 'some', 'interesting', 'things' ]

Testing

You can see examples in the __tests__ folder.

$ npm test
> DEBUG=ava:* nyc ava --color -v

✔ parsingStep › Remove letter A using parsing step
✔ countTermOccurrences › Generate unique terms
✔ tokens › Tokenize documents
✔ classify › Classify a text
✔ uniqueTerms › Generate unique terms

5 tests passed

----------|----------|----------|----------|----------|-------------------|
File      |  % Stmts | % Branch |  % Funcs |  % Lines | Uncovered Line #s |
----------|----------|----------|----------|----------|-------------------|
All files |      100 |      100 |      100 |      100 |                   |
 index.js |      100 |      100 |      100 |      100 |                   |
----------|----------|----------|----------|----------|-------------------|