Overview

jspos is a Javascript port of Mark Watson's FastTag part of speech tagger (itself based on Eric Brill's work). jspos also includes a simple lexer for splitting a text string into words and other taggable tokens.

You can use jspos to take some text like this:

"This is some sample text. This text can contain multiple sentences."

and get back tagged words like this:

This /DT is /VBZ some /DT sample /NN text /NN . /. This /DT text /NN can /MD contain /VB multiple /JJ sentences /NNS . /.

Sample Code

var words = new Lexer().lex("This is some sample text. This text can contain multiple sentences."); var taggedWords = new POSTagger().tag(words); for (i in taggedWords) { var taggedWord = taggedWords[i]; var word = taggedWord[0]; var tag = taggedWord[1]; print(word + " /" + tag); // if running in a browser, try document.writeln() instead of print() }

jspos should be fast enough for many uses - on V8 (Chrome's Javascript Engine), jspos can lex and tag 767 words in around 10 milliseconds (on my MacBook non-pro).

News

June 13 2011 - Version 0.2 has been released, incorporating a compressed lexicon courtesy of Toby Rahilly.