Title: Identifying Subjective Language
1Identifying Subjective Language
- Janyce Wiebe
- University of Pittsburgh
2Overview
- General area acquire knowledge of evaluative
and speculative language and use it in NLP
applications - Primarily corpus-based work
- Today results of exploratory studies
3Collaborators
- Rebecca Bruce, Vasileios Hatzivassiloglou, Joseph
Phillips - Matthew Bell, Melanie Martin,Theresa Wilson
4Subjectivity Tagging
- Recognizing opinions and evaluations
- (Subjective sentences) as opposed to
- material objectively presented as true
- (Objective sentences)
Banfield 1985, Fludernik 1993, Wiebe 1994, Stein
Wright 1995
5Examples
- At several different levels, its a
fascinating tale. subjective - Bell Industries Inc. increased its quarterly
to 10 cents from 7 cents a share. objective
6Subjectivity
?
Enthused Wonderful! Great product
Complained You Idiot! Terrible product
Speculated Maybe
7Examples
- Strong addressee-oriented negative evaluation
- Recognizing flames (Spertus 1997)
- Personal e-mail filters (Kaufer 2000)
I had in mind your facts, buddy, not hers. Nice
touch. Alleges whenever facts posted are not
in your persona of what is real.
8Examples
- Opinionated, editorial language
- IR, text categorization (Kessler et al. 1997)
- Do the writers purport to be objective?
Look, this is a man who has great numbers. We
stand in awe of the Woodstock generations ability
to be unceasingly fascinated by the subject of
itself.
9Examples
- Belief and speech reports
- Information extraction, summarization,
intellectual attribution (Teufel Moens 2000)
Northwest Airlines settled the remaining
lawsuits, a federal judge said.
The cost of health care is eroding our standard
of living and sapping industrial strength,
complains Walter Maher.
10Other Applications
- Review mining (Terveen et al. 1997)
- Clustering documents by ideology (Sack 1995)
- Style in machine translation and generation (Hovy
1987)
11Potential Subjective Elements
Sap potential subjective element
"The cost of health care is eroding standards
of living and sapping industrial strength,
complains Walter Maher.
Subjective element
12Subjectivity
- Multiple types, sources, and targets
Somehow grown-ups believed that wisdom adhered
to youth.
We stand in awe of the Woodstock generations
ability to be unceasingly fascinated by the
subject of itself.
13Outline
- Data and annotation
- Sentence-level classification
- Individual words
- Collocations
- Combinations
14Annotations
Manually tagged existing annotations
Three levels expression level
sentence level document level
15Expression Level Annotations
Perhaps youll forgive me for
reposting his response They promised e 2 yet
more for e 3 really good e? 1 stuff
16Expression Level Annotations
Probably the most natural level
Difficult for manual and automatic tagging
detailed no predetermined classification unit
To date used for training and bootstrapping
17Document Level Annotations
Manual flames in Newsgroups Existing
opinion pieces in the WSJ editorials, letters
to the editor, arts leisure reviews
to reviews More directly
related to applications, but
18Document Level Annotations
Opinion pieces contain objective sentences and
Non-opinion pieces contain subjective sentences
News reports present reactions (van Dijk 1988)
Critics claim Supporters argue
Editorials contain facts supporting the argument
Reviews contain information about the product
19Document Level Annotations
In a WSJ data set
opinion pieces subj
74 obj 26
non-opinion pieces subj
43 obj 57
20Data in this Talk
Sentence level 1000 WSJ sentences 3 judges
reached good agreement after rounds Used for
training and evaluation
Expression level 1000 WSJ sentences (2J)
462 newsgroup messages (2J) 15413 words (1J)
Single round results promising Used to
generate features, and not for evaluation
21Data in this Talk
Document level Existing opinion-piece
annotations used to generate features
Manually refined classifications used for
evaluation Identified editorials not marked
as such Only clear instances labeled To
date 1 judge
Distinct from the other data 3 editions, each
more than 150K words
22Sentence Level Annotations
A sentence is labeled subjective if any
significant expression of subjectivity appears
The cost of health care is eroding our standard
of living and sapping industrial strength,
complains Walter Maher. What an idiot,
the idiot presumably complained.
23Sentence Classification
Probabilistic classifier
Binary Features pronoun, adjective,
number, modal will , adverb not,
new paragraph Lexical feature good for subj
good for obj good for neither
10-fold cross validation 51 baseline 72
average accuracy across folds 82 average
accuracy on sentences rated certain
24Identifying PSEs
There are few high precision, high
frequency potential subjective elements
25Identifying Individual PSEs
- Classifications correlated with adjectives
- Good subsets
- Dynamic adjectives (Quirk et al. 1985)
- Positive, negative polarity gradability
- automatically identified in corpora
- (Hatzivassiloglou McKeown 1997)
-
Results from distributional similarity
26Distributional Similarity
Word similarity based on distributional pattern
of words
Much work in NLP (see Lee 99, Lee and Pereira 99)
- Purposes
- Improve estimates of unseen events
- Thesaurus and dictionary construction from
corpora
27Lins Distributional Similarity
Word R W I R1
have have R2 dog brown R3 dog
. . .
Lin 1998
28Lins Distributional Similarity
Sum over RWint I(Word1,RWint) I(Word2,RWint)
/ Sum over RWw1 I(Word1,RWw1) Sum over RWw2
I(Word2,RWw2)
29Bizarre
strange similar scary unusual
fascinating interesting curious tragic
different contradictory peculiar silly sad
absurd poignant crazy funny comic
compelling odd
30Bizarre
strange similar scary unusual
fascinating interesting curious tragic
different contradictory peculiar silly sad
absurd poignant crazy funny comic
compelling odd
31Bizarre
strange similar scary unusual
fascinating interesting curious tragic
different contradictory peculiar silly sad
absurd poignant crazy funny comic
compelling odd
32Filtering
Filtered Set
Seed Words
Word cluster removed if precision on training
set Words Clusters
33Parameters
Threshold
Seed Words
Words Clusters
Cluster size
34Seeds from Annotations
- 1000 WSJ sentences with sentence level and
expression level annotations - They promised e 2 yet more for
- e 3 really good e? 1 stuff.
- "It's e? 3 really e- 3 bizarre," says Albert
Lerman, creative director at the Wells agency.
35Experiments
1/10 used for training, 9/10 for testing
Parameters Cluster-size fixed at 20
Filtering threshold precision of baseline
adjective feature on the training data
7.5 ave 10-fold cross validation
More improvements with other adj features
36Opinion Pieces
- 3 WSJ data sets, over 150K words each
For measuring precision Prec(S)
instances of S in opinions /
total instances of S
Baseline for comparison words in opinions
/ total words
Skewed distribution 13-17 words in opinions
37Parameters
Threshold
1-70
Seed Words
Words Clusters
2-40
Cluster size
38Results
Varies with parameter settings, but there are
smooth regions of the space
Here training/validation/testing
39Low Frequency Words
Single instance in a corpus low
frequency Analysis of expression level
annotations there are many more
single-instance words in subjective
elements than outside them
40Unique Words
Replace all words that appear once in the test
data with UNIQUE
5-10 points
41Collocations
here we go again get out of here what
a well and good rocket science for
the last time just as well ! Start
with the observation that low precision
words often compose higher precision collocations
42Collocations
Identify n-gram PSEs as sequences whose
precision is higher than the maximum precision of
its constituents
W1,W2 is a PSE if prec(W1,W2) max
(prec(W1),prec(W2))
W1,W2,W3 is a PSE if prec(W1,W2,W3)
max(prec(W1,W2),prec(W3)) or prec(W1,W2,W3)
max(prec(W1),prec(W2,W3))
43Collocations
Moderate improvements 3-10 points
But with all unique words mapped to
UNIQUE 13-24 points
44Example Collocations with Unique
highlyadverb UNIQUEadj highly
unsatisfactory highly unorthodox
highly talented highly conjectural
highly erotic
45Example Collocations with Unique
UNIQUEverb outIN farm out
chuck out ruling out crowd out
flesh out blot out
spoken out luck out
46Collocations
UNIQUEadj toTO UNIQUEverb
impervious to reason strange to
celebrate wise to temper
theypronoun areverb UNIQUEnoun they
are fools they are noncontenders
UNIQUEnoun ofIN itspronoun sum of
its usurpation of its proprietor
of its
47Opinion Results Summary
Best
Worst baseline 17
baseline 13
prec/freq
prec/freq Adjs 21/373
09/2137 Verbs 16/721
07/3193 2-grams 10/569
04/525 3-grams 07/156
03/148 1-U-grams 10/6065
06/6045 2-U-grams 24/294
14/288 3-U-grams 27/138
13/144
Disparate features have consistent performance N
Collocation sets largely distinct
48Does it add up?
Good preliminary results classifying opinion
pieces using density and feature count features.
49Future Work
- Mutual bootstrapping (Riloff Jones 1999)
- Co-training (Collins Singer 1999) to learn both
PSEs and contextual features - Integration into a probabilistic model
- Text classification and review mining
50References
- Banfield, A. (1982). Unspeakable Sentences.
Routledge and Kegan Paul. - Collins, M. Singer, Y. (1999). Unsupervised
models for named entity classification.
EMNLP-VLC-99. - van Dijk, T.A. (1988). News as Discourse.
Lawrence Erlbaum. - Fludernik, M. (1983). The Fictions of Language
and the Languages of Fiction. Routledge. - Hovy, E. (1987). Generating Natural Language
Under Pragmatic Constraints. PhD dissertation. - Kaufer, D. (2000). Flaming. www.eudora.com
- Kessler, B., Nunberg, G., Schutze H. (1997).
Automatic Detection of Genre. ACL-EACL-97. - Riloff, E. Jones R. (1999). Learning
Dictionaries for Information Extraction by
Multi-level Boot-strapping. AAAI-99
51References
- Stein, D. Wright, S. (1995). Subjectivity and
Subjectivisation. Cambridge. - Terveen, W., Hill, W., Amento, B. ,McDonald D.
Creter, J. (1997). Building Task-Specific
Interfaces to High Volume Conversational Data.
CHI-97. - Teufel S., Moens M. (2000). Whats Yours and
Whats Mine Determining Intellectual
Attribution in Scientific Texts. EMNLP-VLC-00. - Wiebe, J. (2000). Learning Subjective Adjectives
from Corpora. AAAI-00. - Wiebe, J. (1994). Tracking Point of View in
Narrative. Computational Linguistics (20) 2. - Wiebe, J. , Bruce, R., OHara T. (1999).
Development and Use of a Gold Standard Data Set
for Subjectivity Classifications. ACL-99.
52References
- Hatzivassiloglou V. McKeown K. (1997).
Predicting the Semantic Orientation of
Adjectives. ACL-EACL-97. - Hatzovassiloglou V. Wiebe J. (2000). Effects of
Adjective Orientation and Gradability on Sentence
Subjectivity. COLING-00. - Lee, L. (1999). Measures of Distributional
Similarity. ACL-99. - Lee, L. Pereira F. (1999). ACL-99.
- Lin, D. (1998). Automatic Retrieval and
Clustering of Similar Words. COLING-ACL-98. - Quirk, R, Greenbaum, S., Leech, G., Svartvik,
J. (1985). A Comprehensive Grammar of the
English Language. Longman. - Sack, W. (1995). Representing and Recognizing
Point of View. AAAI Fall Symposium on Knowledge
Navigation and Retrieval.
53Sentence Annotations
- Ave pair-wise Kappa scores
- all data .69
- certain data .88 (60 of the corpus)
- Case study of analyzing and improving intercoder
- reliability
- if there is symmetric disagreement resulting
from bias - assessed by fitting probability models (Bishop et
al. 1975, CoCo) - bias marginal homogeneity
- symmetric disagreement quasi-symmetry
- use the latent class model to correct
disagreements -
-
-
54Test for Bias Marginal Homogeneity
1 X1
2 X2
3 X3
4 X4
Worse the fit, greater the bias
55Test for Symmetric Disagreement Quasi-Symmetry
Tests relationships among the off-diagonal counts
Better the fit, higher the correlation
56(Potential) Subjective Elements
- Same word, different types
- Great majority objective
- Great! positive evaluative
- Just great. negative evaluative
57Review Mining
From HoodoohoodooBUGZAPPER_at_newnorth.net Newsgro
ups rec.gardens Subject Re Garden software I
bought a copy of Garden Encyclopedia from
Sierra. Well worth the time and money.