en-pos

A better English POS tagger written in JavaScript

Installation and usage

Install via NPM:

npm i --save en-pos

How to use

constTag=require("en-pos").Tag;vartags=newTag(["this","is","my","sentence"]).initial()// initial dictionary and pattern based tagging.smooth()// further context based smoothing.tags;console.log(tags);// ["DT","VBZ","PRP$","NN"]

Annotation Specification

Annotation	Name	Example
NN	Noun	dogman
NNS	Plural noun	dogsmen
NNP	Proper noun	LondonAlex
NNPS	Plural proper noun	Smiths
VB	Base form verb	be
VBP	Present form verb	throw
VBZ	Present form (3rd person)	throws
VBG	Gerund form verb	throwing
VBD	Past tense verb	threw
VBN	Past participle verb	thrown
MD	Modal verb	canshallwillmaymustought
JJ	Adjective	bigfast
JJR	Comparative adjective	bigger
JJS	Superlative adjective	biggest
RB	Adverb	notquicklyclosely
RBR	Comparative adverb	less-closelyfaster
RBS	Superlative adverb	fastest
DT	Determiner	theasomeboth
PDT	Predeterminer	allquite
PRP	Personal Pronoun	Iyouheshe
PRP$	Possessive Pronoun	Iyouheshe
POS	Possessive ending	's
IN	Preposition	ofbyin
PR	Particle	upoff
TO	to	to
WDT	Wh-determiner	whichthatwhateverwhichever
WP	Wh-pronoun	whowhoeverwhomwhat
WP$	Wh-possessive	whose
WRB	Wh-adverb	howwhere
EX	Expletive there	there
CC	Coordinating conjugation	&andnoror
CD	Cardinal Numbers	1777one
LS	List item marker	1BCOne
UH	Interjection	ahohoops
FW	Foreign Words	vivamontoujours
,	Comma	,
:	Mid-sent punct	:;...
.	Sent-final punct.	.!?
(	Left parenthesis	)}]
)	Right parenthesis	({[
#	Pound sign	#
$	Currency symbols	$€£¥
SYM	Other symbols	+*/<>
EM	Emojis & emoticons	:)❤

Accuracy and performance

TL:DR;

• When smoothing is enabled: 96.43% accuracy (processing 132K tokens in 38 seconds)
• When smoothing is disabled: 94.4% accuracy (processing 132K tokens in 3 seconds)

---

As of 25 Jan 2017, this library scored 96.43% at the Penn Treebank test (0.3% away from being a state of the art tagger).

Being written in JavaScript, I think it's safe to say that this is the most accurate JavaScript POS tagger, since the only JS library I know of is pos-js which when I tested on the same treebank scored 87.8%, though it was faster than my implementation when smoothing is enabled.

However, if performance is what's you're after rather than accuracy, then you have the option to disable smoothing in this library and this will marginally increase performance making this library even faster than pos-js but with far better accuracy (94.4%).

Building from source and testing

• Build: tsc (requires typescript)
• Test: node test/test.ts

Credits

• This project is an optimization and (almost complete) re-writing of Compendium's POS tagger.
• Compendium's tagger itself was based on fasttag.
• Fasttag is based on Eric Brill's POS tagger.