POS Tagging: Introduction

Heng Ji
[email protected]
Sept 13, 2010

1
Assignment 1

 Questions?

2/39
Outline
 Parts of speech (POS)
 Tagsets
 POS Tagging
 Rule-based tagging
 Markup Format
 Open source Toolkits

3/39
What is Part-of-Speech (POS)

 Generally speaking, Word Classes (=POS) :

 Verb, Noun, Adjective, Adverb, Article, …
 We can also include inflection:
 Verbs: Tense, number, …
 Nouns: Number, proper/common, …
 Adjectives: comparative, superlative, …
 …

4/39
Parts of Speech
 8 (ish) traditional parts of speech
 Noun, verb, adjective, preposition, adverb, arti
cle, interjection, pronoun, conjunction, etc
 Called: parts-of-speech, lexical categories, wo
rd classes, morphological classes, lexical tag
s...
 Lots of debate within linguistics about the num
ber, nature, and universality of these
 We’ll completely ignore this debate.

5/39
7 Traditional POS Categories

N noun chair, bandwidth, pacing

V verb study, debate, munch
 ADJ adj purple, tall, ridiculous
 ADV adverb unfortunately, slowly,
P preposition of, by, to
 PRO pronoun I, me, mine
 DET determiner the, a, that, those

6/39
POS Tagging
 The process of assigning a part-of-speech or
lexical class marker to each word in a
WORD tag
collection.
the DET
koala N
put V
the DET
keys N
on P
the DET
table N

7/39
Penn TreeBank POS Tag Set

 Penn Treebank: hand-annotated corpus of

Wall Street Journal, 1M words
 46 tags
 Some particularities:
 to /TO not disambiguated
 Auxiliaries and verbs not distinguished

8/39
Penn Treebank Tagset

9/39
 Why POS tagging is useful?
 Speech synthesis:
 How to pronounce “lead”?
 INsult inSULT
 OBject obJECT
 OVERflow overFLOW
 DIScount disCOUNT
 CONtent conTENT
 Stemming for information retrieval
 Can search for “aardvarks” get “aardvark”
 Parsing and speech recognition and etc
 Possessive pronouns (my, your, her) followed by nouns
 Personal pronouns (I, you, he) likely to be followed by verbs
 Need to know if a word is an N or V before you can parse
 Information extraction
 Finding names, relations, etc.
 Machine Translation
10/39
Equivalent Problem in Bioinformatics
 Durbin et al. Biological
Sequence Analysis, Cambridge
University Press.
 Several applications, e.g.
proteins
 From primary structure
ATCPLELLLD
 Infer secondary structure
HHHBBBBBC..

11/39
 Why is POS Tagging Useful?
 First step of a vast number of practical tasks
 Speech synthesis
 How to pronounce “lead”?
 INsult inSULT
 OBject obJECT
 OVERflow overFLOW
 DIScount disCOUNT
 CONtent conTENT

 Parsing
 Need to know if a word is an N or V before you can parse
 Information extraction
 Finding names, relations, etc.
 Machine Translation
12/39
Open and Closed Classes
 Closed class: a small fixed membership
 Prepositions: of, in, by, …
 Auxiliaries: may, can, will had, been, …
 Pronouns: I, you, she, mine, his, them, …
 Usually function words (short common words which
play a role in grammar)
 Open class: new ones can be created all the time
 English has 4: Nouns, Verbs, Adjectives, Adverbs
 Many languages have these 4, but not all!

13/39
 Open Class Words
 Nouns
 Proper nouns (Boulder, Granby, Eli Manning)
 English capitalizes these.
 Common nouns (the rest).
 Count nouns and mass nouns
 Count: have plurals, get counted: goat/goats, one goat, two goats
 Mass: don’t get counted (snow, salt, communism) (*two snows)
 Adverbs: tend to modify things
 Unfortunately, John walked home extremely slowly yesterday
 Directional/locative adverbs (here,home, downhill)
 Degree adverbs (extremely, very, somewhat)
 Manner adverbs (slowly, slinkily, delicately)
 Verbs
 In English, have morphological affixes (eat/eats/eaten)

14/39
Closed Class Words
Examples:
 prepositions: on, under, over, …
 particles: up, down, on, off, …
 determiners: a, an, the, …
 pronouns: she, who, I, ..
 conjunctions: and, but, or, …
 auxiliary verbs: can, may should, …
 numerals: one, two, three, third, …

15/39
Prepositions from CELEX

16/39
English Particles

17/39
Conjunctions

18/39
POS Tagging
Choosing a Tagset

 There are so many parts of speech, potential distinctions we ca

n draw
 To do POS tagging, we need to choose a standard set of tags to
work with
 Could pick very coarse tagsets
 N, V, Adj, Adv.

 More commonly used set is finer grained, the “Penn TreeBank t

agset”, 45 tags
 PRP$, WRB, WP$, VBG

 Even more fine-grained tagsets exist

19/39
Using the Penn Tagset
 The/DT grand/JJ jury/NN commmented/VBD
on/IN a/DT number/NN of/IN other/JJ topics/N
NS ./.
 Prepositions and subordinating conjunctions
marked IN (“although/IN I/PRP..”)
 Except the preposition/complementizer “to” is
just marked “TO”.

20/39
POS Tagging
 Words often have more than one POS: back
 The back door = JJ
 On my back = NN
 Win the voters back = RB
 Promised to back the bill = VB
 The POS tagging problem is to determine the
POS tag for a particular instance of a word.

These examples from Dekang Lin

21/39
How Hard is POS Tagging?
Measuring Ambiguity

22/39
Current Performance

 How many tags are correct?

 About 97% currently
 But baseline is already 90%
 Baseline algorithm:
 Tag every word with its most frequent tag
 Tag unknown words as nouns
 How well do people do?

23/39
Quick Test: Agreement?

 the students went to class

 plays well with others
 fruit flies like a banana DT: the, this, that
NN: noun
VB: verb
P: prepostion
ADV: adverb

24/39
 How to do it? History
DeRose/Church Trigram Tagger Combined Methods
Efficient HMM (Kempe) 98%+
Sparse Data 96%+
95%+ Tree-Based Statistics
(Helmut Shmid)
Transformation
Rule Based – 96%+
Greene and Rubin HMM Tagging Based Tagging
Rule Based - 70% (CLAWS) (Eric Brill)
Rule Based – 95%+ Neural Network
93%-95% 96%+

1960 1970 1980 1990 2000

Brown Corpus Brown Corpus LOB Corpus

Created (EN-US) Tagged Tagged
1 Million Words
British National
POS Tagging
Corpus
separated from
LOB Corpus (tagged by CLAWS)
other NLP
Created (EN-UK)
1 Million Words Penn Treebank
Corpus
(WSJ, 4.5M)

25/39
Two Methods for POS Tagging
1. Rule-based tagging
 (ENGTWOL)
2. Stochastic
1. Probabilistic sequence models
 HMM (Hidden Markov Model) tagging
 MEMMs (Maximum Entropy Markov Models)

26/39
Rule-Based Tagging
 Start with a dictionary
 Assign all possible tags to words from the
dictionary
 Write rules by hand to selectively remove
tags
 Leaving the correct tag for each word.

27/39
Rule-based taggers
 Early POS taggers all hand-coded
 Most of these (Harris, 1962; Greene and Rubin, 197
1) and the best of the recent ones, ENGTWOL (Voutil
ainen, 1995) based on a two-stage architecture
 Stage 1: look up word in lexicon to give list of potential
POSs
 Stage 2: Apply rules which certify or disallow tag
sequences
 Rules originally handwritten; more recently Machine
Learning methods can be used

28/39
 Start With a Dictionary
• she: PRP
• promised: VBN,VBD
• to TO
• back: VB, JJ, RB, NN
• the: DT
• bill: NN, VB

• Etc… for the ~100,000 words of English with more than 1

tag

29/39
Assign Every Possible Tag

NN
RB
VBN JJ VB
PRP VBD TO VB DT NN
She promised to back the bill

30/39
Write Rules to Eliminate Tags
Eliminate VBN if VBD is an option when VBN|
VBD follows “<start> PRP”
NN
RB
VBN JJ VB
PRP VBD TO VB DT NN
She promised to back the bill

31/39
Inline Mark-up
 POS Tagging
http://nlp.cs.qc.cuny.edu/wsj_pos.zip
 Input Format
Pierre Vinken, 61/CD years/NNS old , will join th
e board as a nonexecutive director Nov. 29.
 Output Format
Pierre/NNP Vinken/NNP ,/, 61/CD years/NNS old/
JJ ,/, will/MD join/VB the/DT board/NN as/IN a/
DT nonexecutive/JJ director/NN Nov./NNP 29/
CD ./.

32/39
POS Tagging Tools
 Stanford tagger (Loglinear tagger )
http://nlp.stanford.edu/software/tagger.shtml
 Brill tagger
 http://www.tech.plym.ac.uk/soc/staff/guidbugm/software/RULE_B
ASED_TAGGER_V.1.14.tar.Z
 tagger LEXICON test BIGRAMS LEXICALRULEFULE CONTEXTU
ALRULEFILE
 YamCha (SVM)
http://chasen.org/~taku/software/yamcha/
 MXPOST (Maximum Entropy)
ftp://ftp.cis.upenn.edu/pub/adwait/jmx/
 More complete list at:
http://www-nlp.stanford.edu/links/statnlp.html#Taggers

33/39
NLP Toolkits
 Uniform CL Annotation Platform
 UIMA (IBM NLP platform): http://incubator.apache.org/uima/svn.html
 Mallet (UMASS): http://mallet.cs.umass.edu/index.php/Main_Page
 MinorThird (CMU): http://minorthird.sourceforge.net/
 NLTK: http://nltk.sourceforge.net/
Natural langauge toolkit, with data sets  Demo

 Information Extraction
 Jet (NYU IE toolkit) http://www.cs.nyu.edu/cs/faculty/grishman/jet/license.ht
ml
 Gate: http://gate.ac.uk/download/index.html
University of Sheffield IE toolkit
 Information Retrieval
 INDRI: http://www.lemurproject.org/indri/
Information Retrieval toolkit
 Machine Translation
 Compara: http://adamastor.linguateca.pt/COMPARA/Welcome.html
 ISI decoder: http://www.isi.edu/licensed-sw/rewrite-decoder/
 MOSES: http://www.statmt.org/moses/

34/39
Looking Ahead: Next Class

 Machine Learning for POS Tagging:

Hidden Markov Model

35/39