retext-porter-stemmer

retext-porter-stemmer

retext implementation of the Porter stemming algorithm.

Installation

npm:

$ npm install retext-porter-stemmer

Component.js:

$ component install wooorm/retext-porter-stemmer

Bower:

$ bower install retext-porter-stemmer

Duo:

var porterStemmer = require('wooorm/retext-porter-stemmer');

Usage

var Retext = require('retext');
var visit = require('retext-visit');
var inspect = require('retext-inspect');
var porterStemmer = require('retext-porter-stemmer');

var retext = new Retext()
    .use(inspect)
    .use(visit)
    .use(porterStemmer)

retext.parse('A simple english sentence.', function (err, tree) {
    tree.visit(tree.WORD_NODE, function (node) {
        console.log(node);
    });
    /**
     * WordNode[1] [data={"stem":"a"}]
     * └─ TextNode: 'A'
     * WordNode[1] [data={"stem":"simpl"}]
     * └─ TextNode: 'simple'
     * WordNode[1] [data={"stem":"english"}]
     * └─ TextNode: 'english'
     * WordNode[1] [data={"stem":"sentenc"}]
     * └─ TextNode: 'sentence'
     */
});

API

None, retext-porter-stemmer automatically detects the stem of each WordNode (using wooorm/stemmer), and stores the stem in node.data.stem.

Performance

On a MacBook Air, retext performs about 11% slower with retext-porter-stemmer.

           retext w/o retext-porter-stemmer
  221 op/s » A paragraph (5 sentences, 100 words)
   24 op/s » A section (10 paragraphs, 50 sentences, 1,000 words)

           retext w/ retext-porter-stemmer
  197 op/s » A paragraph (5 sentences, 100 words)
   20 op/s » A section (10 paragraphs, 50 sentences, 1,000 words)

Related

• retext-lancaster-stemmer — Same workings, but uses the Lancaster (Paice/Husk) stemming algorithm.

License

MIT © Titus Wormer