s ubjectivity and s entiment a nalysis : from w ords to d iscourse jan wiebe department of computer...
TRANSCRIPT
![Page 1: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/1.jpg)
SUBJECTIVITY AND SENTIMENT ANALYSIS: FROM WORDS TO DISCOURSE
Jan Wiebe
Department of Computer ScienceIntelligent Systems Program
University of Pittsburgh
![Page 2: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/2.jpg)
BURGEONING FIELD
Quite a large problem space Several terms reflecting varying goals
and modelsSentiment AnalysisOpinion Mining Opinion ExtractionSubjectivity AnalysisAppraisal AnalysisAffect SensingEmotion Detection Identifying PerspectiveEtc.
![Page 3: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/3.jpg)
WHAT IS SUBJECTIVITY?
The linguistic expression of somebody’s opinions, sentiments, emotions, evaluations, beliefs, speculations (private states)
Private state: state that is not open to objective observation or verification Quirk, Greenbaum, Leech, Svartvik (1985).
Note that this particular use of subjectivity is adaptedfrom literary theory E.G. Banfield 1982, Fludernik 1993; Wiebe PhD Dissertation 1990.
![Page 4: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/4.jpg)
EXAMPLES OF SUBJECTIVE EXPRESSIONS
References to private statesShe was enthusiastic about the planHe was boiling with anger
References to speech or writing events expressing private statesLeaders rounding condemned his verbal
assault on Israel Expressive subjective elements
That would lead to disastrous consequences
What a freak show
![Page 5: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/5.jpg)
direct subjective span: are happy source: <writer, I, People> attitude:
inferred attitude span: are happy because Chavez has fallen type: neg sentiment intensity: medium target:
target span: Chavez has fallen
target span: Chavez
attitude span: are happy type: pos sentiment intensity: medium target:
direct subjective span: think source: <writer, I> attitude:
attitude span: think type: positive arguing intensity: medium target:
target span: people are happy because Chavez has fallen
I think people are happy because Chavez has fallen
MPQA corpus: http://www.cs.pitt.edu/mpqa
MANUALLY (HUMAN) ANNOTATED NEWS DATA WILSON PHD DISSERTATION 2008
![Page 6: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/6.jpg)
FOCUS
Our focus is linguistic disambiguation; how should language be interpreted? Is it subjective in the first place? If so, is it
positive or negative? What is it about? Etc. Subjective language is highly ambiguous
![Page 7: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/7.jpg)
INTERPRETATION
Lexicon ofkeywords out of context
Full contextualInterpretationof words in text or dialogue
continuum
“The dream”NLP methods/resourcesbuilding toward fullinterpretations
Today: several tasks along the continuum
![Page 8: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/8.jpg)
INTERPRETATION
Lexicon ofkeywords out of context
Full contextualInterpretationof words in textor dialogue
continuum
BrilliantDifferenceHateInterestLove…
![Page 9: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/9.jpg)
SUBJECTIVITY LEXICONS
Most approaches to subjectivity and sentiment analysis exploit subjectivity lexicons. Lists of keywords that have been gathered
together because they have subjective uses
![Page 10: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/10.jpg)
AUTOMATICALLY IDENTIFYING SUBJECTIVE WORDS
Much work in this area
E.g. Hatzivassiloglou & McKeown 1997; Wiebe 2000; Turney 2002; Kamps & Marx 2002; Wiebe, Riloff, Wilson 2003; Kim & Hovy 2005; Esuli & Sebastiani 2005; Mohammad, Dorr, Dunne 2009
Subjectivity Lexicon: http://www.cs.pitt.edu/mpqa
Entries from several sources (our work and others’)
![Page 11: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/11.jpg)
HOWEVER…
Consider the keyword “Interest”. It is in the subjectivity lexicon. But, what about “interest rate”, for example?
![Page 12: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/12.jpg)
DICTIONARY DEFINITIONS SENSES
Interest, involvement -- (a sense of
concern with and curiosity about someone or something; "an interest in music")
Interest -- (a fixed charge for
borrowing money; usually a percentage of the amount borrowed; "how much interest do you pay on your mortgage?")
![Page 13: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/13.jpg)
DICTIONARY DEFINITIONS SENSES
Interest, involvement -- (a sense of
concern with and curiosity about someone or something; "an interest in music")
Interest -- (a fixed charge for
borrowing money; usually a percentage of the amount borrowed; "how much interest do you pay on your mortgage?")
S
O
![Page 14: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/14.jpg)
SENSES
Even in subjectivity lexicons, many senses of the keywords are objective ~50% in our study!
Thus, many appearances of keywords in texts are false hits
![Page 15: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/15.jpg)
SENSES
His alarm grew as the election returns came in.
He set his alarm for 7am.
His trust grew as the candidate spoke. His trust grew as interest rates
increased.
![Page 16: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/16.jpg)
WORDNET MILLER 1995; FELLBAUM 1998
![Page 17: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/17.jpg)
EXAMPLES
“There are many differences between African and Asian elephants.”
“… dividing by the absolute value of the difference from the mean…”
“Their differences only grew as they spent more time together …”
“Her support really made a difference in my life”
“The difference after subtracting X from Y…”
![Page 18: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/18.jpg)
SUBJECTIVITY SENSE LABELING
Automatically classifying senses as subjective or objective
Wiebe & Mihalcea 2006Gyamfi, Wiebe, Mihalcea, Akkaya 2009
See also: Esuli & Sebastiani 2006, 2007 Andreevskaia & Bergler 2006a,b Su & Markert 2008,2009
![Page 19: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/19.jpg)
INTERPRETATION
Lexicon of keywords out of context
Full contextualInterpretationof words in text or dialog
continuum
Brilliant sense#1 S sense#2 S …Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O…
Now we will leave the lexicon and look at disambiguation in the context of a text orconversation
![Page 20: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/20.jpg)
SubjectivitySentenceClassifier
CONTEXTUAL SUBJECTIVITY ANALYSIS
“He spins a riveting plot which grabs and holds the reader’s interest…”
S O?
S O?
“The notes do not pay interest.”
Do the sentences contain subjectivity?
E.g. Riloff & Wiebe 2003 Yu & Hatzivassiloglou 2003
![Page 21: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/21.jpg)
SubjectivityPhrase
Classifier
CONTEXTUAL SUBJECTIVITY ANALYSIS
“He spins a riveting plot which grabs and holds the reader’s interest…”
S O?
S O?
“The notes do not pay interest.”
Is a phrase containing a keyword subjective?
Wilson, Wiebe, Hoffmann 2005
![Page 22: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/22.jpg)
CONTEXTUAL SUBJECTIVITY ANALYSIS
S O?
S O?
Is a phrase containing a keyword positive,Negative, or neutral?
Wilson, Wiebe, Hoffmann 2005
SentimentPhrase
Classifier
Pos, Neg, Neutral?
Pos, Neg, Neutral?
“There are many differences between African and Asian elephants.”
“Their differences only grew as they spent more time together …”
We’ll return to this, topic after next.But first…
![Page 23: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/23.jpg)
INTERPRETATION
Lexicon of keywords out of context
Full contextualInterpretationof words in text or dialog
continuum
Brilliant sense#1 S sense#2 S …Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O…
ContextualSubjectivityanalysis
Exploiting sense labels to improvethe contextual classifiers
![Page 24: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/24.jpg)
SubjectivityClassifier
S Sense 4 “a sense of concern with and curiosity about someone or something”
O Sense 1 “a fixed charge for borrowing money”
SUBJECTIVITY TAGGING USING WSD
“The notes do not pay interest.”
“He spins a riveting plot which grabs and holds the reader’s interest…”
WSDSystem
Sense 4
Sense 1
S O?
S O?
![Page 25: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/25.jpg)
SubjectivityClassifier
S Sense 4 “a sense of concern with and curiosity about someone or something”
O Sense 1 “a fixed charge for borrowing money”
SUBJECTIVITY TAGGING USING WSD
“The notes do not pay interest.”
“He spins a riveting plot which grabs and holds the reader’s interest…”
WSDSystem
Sense 4
Sense 1S O
S O
![Page 26: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/26.jpg)
26
EXAMPLES
“There are many differences between African and Asian elephants.” Sense#1 O
“… dividing by the absolute value of the difference from the mean…” Sense#2 O
“Their differences only grew as they spent more time together …” Sense#3 S
“Her support really made a difference in my life” Sense#4 S
“The difference after subtracting X from Y…” Sense#5 O
![Page 27: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/27.jpg)
27
“There are many differences between African and Asian elephants.” Sense#1 O
“… dividing by the absolute value of the difference from the mean…” Sense#2 O
“Their differences only grew as they spent more time together …” Sense#3 S
“Her support really made a difference in my life” Sense#4 S
“The difference after subtracting X from Y…” Sense#5 O
Is is one of these?
![Page 28: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/28.jpg)
28
“There are many differences between African and Asian elephants.” Sense#1 O
“… dividing by the absolute value of the difference from the mean…” Sense#2 O
“Their differences only grew as they spent more time together …” Sense#3 S
“Her support really made a difference in my life” Sense#4 S
“The difference after subtracting X from Y…” Sense#5 O
![Page 29: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/29.jpg)
SubjectivityClassifier
SUBJECTIVITY TAGGING USING SUBJECTIVITY WSD
SWSDSystem
S O?
Sense O {1, 2, 5}
Sense S {3,4}
S O?
Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O
“There are many differences between African and Asian elephants.”
“Their differences only grew as they spent more time together …”
![Page 30: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/30.jpg)
SubjectivityClassifier
SUBJECTIVITY TAGGING USING SUBJECTIVITY WSD
SWSDSystem
S O
Sense O {1, 2, 5}
Sense S {3,4}
S O
Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O
“There are many differences between African and Asian elephants.”
“Their differences only grew as they spent more time together …”
![Page 31: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/31.jpg)
SWSD AKKAYA, WIEBE, MIHALCEA 2009 AKKAYA, CONRAD, WIEBE, MIHALCEA 2010 AKKAYA, WIEBE, CONRAD MIHALCEA 2011
Compared system performance when WSD: Using the full sense inventory SWSD: Using only two senses, subj-sense and obj-
sense SWSD Performance is well above baseline and
the performance of full WSD SWSD is a feasible variant of WSD Subjectivity provides a natural course-grained sense
grouping Two types of data:
SENSEVAL data with senses mapped to S/O senses Data acquired using Amazon Mechanical Turk
Workers shown a target word in a sentence and two sets of senses (the S and O sets). Task: which set matches the EG?
![Page 32: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/32.jpg)
SWSD IN SUBJECTIVITY TAGGING
SWSD exploited to improve performance of subjectivity analysis systems
Both S/O and Pos/Neg/Neutral classifiers
![Page 33: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/33.jpg)
SENTIMENT ANALYSIS USING SWSD
SWSDSystem
Sense O {1, 2, 5}
Sense S {3,4}
Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O
“There are many differences between African and Asian elephants.”
“Their differences only grew as they spent more time together …”
Sentiment Classifier
Pos, Neg, Neutral?
Pos, Neg, Neutral?
![Page 34: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/34.jpg)
INTERPRETATION
Lexicon of keywords out of context
Full contextualInterpretationof words in text or dialog
continuum
Brilliant sense#1 S sense#2 S …Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O…
SWSD ContextualSentimentAnalysis
Rest of the talk: contextual processing not boundto word senses
Return to contextual sentiment classification
![Page 35: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/35.jpg)
SENTIMENT ANALYSIS WILSON, WIEBE, HOFFMAN 2005, 2009
Automatically identifying positive and negative emotions, evaluations, and stances Our approach: classify expressions containing a
keyword as positive, negative, both, or neutral
![Page 36: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/36.jpg)
PHRASE-LEVEL SENTIMENT ANALYSIS
See also, E.G. Yi, Nasukawa, Bunescu, Niblack 2003; Polanyi & Zaenen 2004; Popescu & Etzioni 2005; Suzuki, Takamura, Okumura 2006; Moilanen & Pulman 2007; Choi & Cardie 2008
![Page 37: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/37.jpg)
PRIOR VERSUS CONTEXTUAL POLARITY
Many subjectivity lexicons contain polarity information
Prior polarity: out of context, positive, negative, or neutral
A word may appear in a phrase that expresses a different polarity in context
Contextual polarity
![Page 38: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/38.jpg)
MPQA (HUMAN) POLARITY ANNOTATIONS
Judge the contextual polarity of the sentiment that is ultimately being conveyed in the context of the text or conversation
![Page 39: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/39.jpg)
CONTEXTUAL INTERPRETATION
They have not succeeded, and will never succeed, in breaking the will of this valiant people.
![Page 40: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/40.jpg)
CONTEXTUAL INTERPRETATION
They have not succeeded, and will never succeed, in breaking the will of this valiant people.
![Page 41: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/41.jpg)
CONTEXTUAL INTERPRETATION
They have not succeeded, and will never succeed, in breaking the will of this valiant people.
![Page 42: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/42.jpg)
CONTEXTUAL POLARITY IS COMPLEX
They have not succeeded, and will never succeed, in breaking the will of this valiant people.
![Page 43: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/43.jpg)
APPROACH
Step 1: Neutral or Polar? Step 2: Are the polar instances Positive or
Negative? Combine a variety of evidence
![Page 44: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/44.jpg)
EVIDENCE
Modifications and ConjunctionsCheers to Timothy Whitfield for the wonderfully
horrid visuals
Disdain and wrathHatzivassiloglou & McKeown 1997
Subjectivity of the surrounding context; syntactic role in the sentence; etc.
pos
mod
wonderfully horrid
disdain (neg) and wrath(neg)
![Page 45: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/45.jpg)
POLARITY INFLUENCERS
Negation Local not good Longer-distance dependencies
Does not look very good (proposition) No politically prudent Israeli could support either of
them (subject) Phrases with negations may intensify instead
Not only good, but amazing!
![Page 46: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/46.jpg)
POLARITY INFLUENCERS
Contextual Valence Shifters Polanyi & Zaenan 2004 General polarity shifter
Pose little threat Contains little truth
Negative polarity shifters Lack of understanding
Positive polarity shifters Abate the damage
![Page 47: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/47.jpg)
APPROACH
Step 1: Neutral or Polar? Step 2: Are the polar instances Positive or
Negative? Combine a variety of evidence Still much to do in the area of recognizing
contextual polarity
![Page 48: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/48.jpg)
INTERPRETATION
Lexicon of keywords out of context
Full contextualInterpretationof words in text or dialog
continuum
Brilliant sense#1 S sense#2 S …Difference sense#1 O sense#2 O sense#3 S sense#4 S sense#5 O…
SWSD ContextualSentimentAnalysis
Discourse
![Page 49: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/49.jpg)
50
DISCOURSE-LEVEL TREATMENT
Interdependent interpretation of opinions More information about the overall stance
50
Somasundaran & Wiebe 2009; Somasundaran et al. 2009a,b; 2008a,b
See also: Bansal,Cardie,Lee 2008; Thomas,Pang,Lee 2006; Diermeier,Godbout,Yu,Kaufmann 2007; Malouf & Mullen 2008; Lin and Hauptmann 2006; Greene & Resnik 2009; Jiang & Argamon 2008; Klebanov, Diermeier, Beigman 2008; Polanyi & Zaenan 2006; Asher, Benamara, Matheiu 2008; Hirst, Riabinin, Graham 2010
![Page 50: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/50.jpg)
51
MOTIVATION: INTERDEPENDENT INTERPRETATION OF OPINIONS
D::... this kind of rubbery material, it’s a bit more bouncy, like you said they
get chucked around a lot. A bit more durable and that can also be
ergonomic and it kind of feels a bit different from all the other remote
controls.
Example from the AMI Meeting corpus (Carletta et al., 2005)• Scenario-based goal oriented meeting, where the participants
have to design a new TV remote
![Page 51: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/51.jpg)
52
MOTIVATION: INTERDEPENDENT INTERPRETATION OF OPINIONS
D::... this kind of rubbery material, it’s a bit more bouncy, like you said they
get chucked around a lot. A bit more durable and that can also be
ergonomic and it kind of feels a bit different from all the other remote
controls.
positive
positive
positive
?
![Page 52: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/52.jpg)
53
D::... this kind of rubbery material, it’s a bit more bouncy, like you said they
get chucked around a lot. A bit more durable and that can also be
ergonomic and it kind of feels a bit different from all the other remote
controls.
MOTIVATION: INTERDEPENDENT INTERPRETATION OF OPINIONS
positive
positive
positive
?
Observation:1. Speaker is talking about the same thing
![Page 53: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/53.jpg)
54
MOTIVATION: INTERDEPENDENT INTERPRETATION OF OPINIONS
D::... this kind of rubbery material, it’s a bit more bouncy, like you said they
get chucked around a lot. A bit more durable and that can also be
ergonomic and it kind of feels a bit different from all the other remote
controls.
positive
positive
positive
?
Observation:1. Speaker is talking about the same thing2. Speaker is reinforcing his stance (pro-rubbery material)
![Page 54: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/54.jpg)
55
MOTIVATION: INTERDEPENDENT INTERPRETATION OF OPINIONS
D::... this kind of rubbery material, it’s a bit more bouncy, like you said they
get chucked around a lot. A bit more durable and that can also be
ergonomic and it kind of feels a bit different from all the other remote
controls.
positive
positive
positive
Observation:1. Speaker is talking about the same thing2. Speaker is reinforcing his stance (pro-rubbery material) Interpretation coherent with the discourse:Being “a bit different from other remote controls” is positive
positive
Discourse-level relations can help disambiguation of difficult cases
![Page 55: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/55.jpg)
56
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
MOTIVATION: MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
![Page 56: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/56.jpg)
57
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
MOTIVATION:MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
Prediction: Stance regarding the curved shape
QA System: Will the curved shape be accepted?
![Page 57: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/57.jpg)
58
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
Direct opinion
MOTIVATION:MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
![Page 58: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/58.jpg)
59
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
Direct opinionOpinions towards
mutually exclusive option (alternative)
MOTIVATION:MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
![Page 59: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/59.jpg)
60
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
Direct opinionOpinions towards
mutually exclusive option (alternative)
MOTIVATION:MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
![Page 60: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/60.jpg)
61
• Shapes should be curved, so round shapes Nothing square-like.
• ... So we shouldn’t have too square corners and that kind of
thing.
positive
Direct opinionOpinions towards
mutually exclusive option (alternative)
Discourse-level relations can provide More opinion information regarding the
stance
MOTIVATION:MORE INFORMATION ABOUT THE OPINION STANCE
negative
negative
![Page 61: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/61.jpg)
62
THIS WORK
Discourse-level relations
Overall stance classification
Expression-level (fine-grained) Opinion polarity classification
![Page 62: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/62.jpg)
63
THIS WORK
Discourse-level relations
Overall stance classification
Expression-level (fine-grained) Opinion polarity classification
Improve recognition of expression polarity Meeting dataLinguistic SchemeData Annotation Classifiers to recognize individual componentsGlobal inference to model interdependent interpretation of opinions in the discourse
Improve recognition of person’s overall stanceOnline debates and Web dataUnsupervised learning of relevant opinion relationsConcession handling to address specific discourse relations
![Page 63: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/63.jpg)
64
DISCOURSE-LEVEL RELATIONS
Opinion expressions are related in the discourse via
the relation between their targets [what the opinion is about] and whether / how the opinions contribute to an overall stance
![Page 64: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/64.jpg)
65
TARGET RELATIONS
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
positive
positive
positive
negative
![Page 65: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/65.jpg)
66
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
TARGET RELATIONSpositive
positive
positive
negative
same
![Page 66: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/66.jpg)
67
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
TARGET RELATIONSpositive
positive
positive
negative
alternative
same
![Page 67: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/67.jpg)
68
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
DISCOURSE-LEVEL RELATIONS
positive
positive
positive
negative
alternative
same
![Page 68: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/68.jpg)
69
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
DISCOURSE-LEVEL RELATIONS
positive
positive
positive
negative
alternative
samereinforcing
![Page 69: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/69.jpg)
70
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
DISCOURSE-LEVEL RELATIONS
positive
positive
positive
negative
alternative
samereinforcing
![Page 70: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/70.jpg)
71
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
DISCOURSE-LEVEL RELATIONS
positive
positive
positive
negative
alternative
samereinforcing
reinforcing
![Page 71: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/71.jpg)
72
DISCOURSE-LEVEL RELATIONS
The red remote is inexpensive,
but the blue one is cool
The blue remote is cool,
However, it is expensive
positive
positive
positive
negative
alternative
same
non-reinforcing
non-reinforcing
![Page 72: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/72.jpg)
73
This blue remote is cool.
What’s more, the rubbery material is ergonomic.
I feel the red remote is a better choice.
The blue remote will be too expensive.
DISCOURSE-LEVEL RELATIONS
positive
positive
positive
negative
alternative
samereinforcing
reinforcing
<Pos, Pos, same>
<Pos, Neg, alternative>
![Page 73: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/73.jpg)
74
THIS WORK
Discourse-level relations
Overall stance classification
Expression-level (fine-grained) Opinion polarity classification
Improve recognition of expression polarity Meeting dataLinguistic SchemeData Annotation Supervised learning, feature engineeringGlobal inference to model interdependent interpretation of opinions in the discourse
Improve recognition of person’s overall stanceOnline debates and Web dataUnsupervised learning of relevant opinion relationsConcession handling to address specific discourse relations
![Page 74: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/74.jpg)
75
POLARITY TARGET PAIRS
Unsupervised Do not have target and discourse
relations between opinions annotated The data are on-line debates, in which
people largely reinforce their stances Our basic unit is the polarity-target
pair (computed automatically) Mine web data for reinforcing relations
![Page 75: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/75.jpg)
76
This blue remote is cool.
What’s more, the rubbery material is
ergonomic. Blue remote -- positive
rubbery material -- positive
reinforcing
Find via web mining that these support the same stance
![Page 76: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/76.jpg)
77
DATA
Debate: iPhone vs. Blackberry
iPhone of course. Blackberry is now for the senior businessmen market! The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology. The iPhone can reach a very diversified clientele …
![Page 77: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/77.jpg)
78
DATA
Debate: iPhone vs. Blackberry
iPhone of course. Blackberry is now for the senior businessmen market! The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology. The iPhone can reach a very diversified clientele …
Arguing why their stance is correct
![Page 78: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/78.jpg)
79
DATA
Debate: iPhone vs. Blackberry
iPhone of course. Blackberry is now for the senior businessmen market! The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology. The iPhone can reach a very diversified clientele …
Alternatively, justifying why the opposite side is not good
![Page 79: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/79.jpg)
80
DATA
Debate: iPhone vs. Blackberry
iPhone of course. Blackberry is now for the senior businessmen market! The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology. The iPhone can reach a very diversified clientele …
Multiple positive opinions toward the iPhone reinforce a pro-iPhone stanceMultiple negative opinions toward the alternative further reinforce the pro-iPhone stance
Side Classification:
pro-iPhone stance
![Page 80: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/80.jpg)
81
http://www.convinceme.net/
![Page 81: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/81.jpg)
82
http://www.convinceme.net/
Side Classification:
pro-iPhone stanceSide Classification:
pro-Blackberry stance
Side Classification:
pro-iPhone stance
Topics:1. iPhone2. Blackberry
Sides/ Stances:1. Pro-iPhone2. Pro-Blackberry
Dual-topic, Dual-sided debates regarding Named Entities
![Page 82: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/82.jpg)
83
WEB MINING
![Page 83: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/83.jpg)
84
WEB MINING
Stance-1Pro-iPhone
Stance-2Pro-Blackberry
iPhone vs. Blackberry
![Page 84: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/84.jpg)
85
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone +
iPhone vs. Blackberry
![Page 85: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/85.jpg)
86
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry +
iPhone vs. Blackberry
![Page 86: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/86.jpg)
87
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - Blackberry +
Argue for a pro-iPhone stance via negative opinion towards the alternative target (Blackberry)
iPhone vs. Blackberry
![Page 87: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/87.jpg)
88
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Argue for a pro-iPhone stance via negative opinion towards the alternative target (Blackberry)
Argue for a pro-blackberry stance via negative opinion towards the alternative target (iPhone)
iPhone vs. Blackberry
![Page 88: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/88.jpg)
89
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Topic polarity pairs that reinforce a pro-iPhone stance
Topic polarity pairs that reinforce a pro-BB stance
iPhone vs. Blackberry
![Page 89: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/89.jpg)
90
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Target-1 + Target-2 + Target-3 -
post
iPhone vs. Blackberry
![Page 90: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/90.jpg)
91
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Target-1 + Target-2 + Target-3 -
post
iPhone vs. Blackberry
If these all mention the topic, the task is straightforward
![Page 91: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/91.jpg)
92
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Pearl + keyboard + battery -
post
iPhone vs. Blackberry
![Page 92: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/92.jpg)
93
DEBATE TOPICS ARE EVOKED IN A VARIETY OF WAYS
Pro-blackberry
The Pearl does music and video nicely …
First, you still can't beat the full QWERTY keyboard for quick, effortless typing.
Pro-iPhone
Well, Apple has always been a well known company.
Its MAC OS is also a unique thing.
![Page 93: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/93.jpg)
94
Pro-blackberry
The Pearl does music and video nicely …
First, you still can't beat the full QWERTY keyboard for quick, effortless typing.
Pro-iPhone
Well, Apple has always been a well known company.
Its MAC OS is also a unique thing.
Type of Blackberry
Feature of Blackberry
Maker of iPhone
Feature of iPhone
DEBATE TOPICS ARE EVOKED IN A VARIETY OF WAYS
![Page 94: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/94.jpg)
95
Pro-blackberry
The Pearl does music and video nicely …
First, you still can't beat the full QWERTY keyboard for quick, effortless typing.
Pro-iPhone
Well, Apple has always been a well known company.
Its MAC OS is also a unique thing.
Unique Aspects
DEBATE TOPICS ARE EVOKED IN A VARIETY OF WAYS
![Page 95: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/95.jpg)
96
iPhone and Blackberry, both Have e-mail facilities Can be used to take photos Operate on batteries Etc.
Both sides share aspects
SHARED ASPECTS
![Page 96: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/96.jpg)
97
Faster keyboard input
People expressing positive opinions regarding keyboards (generally) prefer Blackberry
SHARED ASPECTS - EXAMPLE
![Page 97: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/97.jpg)
98
Faster keyboard input
Certain shared aspects may be perceived to be better in one side•Keyboards in blackberry
Value for shared aspects depends on personal preferences•Music •KeyboardsPeople argue about what they value
SHARED ASPECTS
![Page 98: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/98.jpg)
99
keyboard+
SHARED ASPECTS
How likely is it to be used to reinforce a pro-iPhone stancepro-Blackberry stance
![Page 99: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/99.jpg)
100
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Pearl + keyboard + battery -
post
iPhone vs. Blackberry
![Page 100: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/100.jpg)
101
WEB MINING
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Pearl + keyboard + battery -
post
iPhone vs. Blackberry
Likelihood of Reinforcement associations
![Page 101: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/101.jpg)
102
ASSOCIATIONS WITH TOPIC-POLARITY
For each opinion-target (targetjp) calculate its
association with each of the opinion-topics P(topic1+|targetj+) P(topic1-|targetj+) P(topic2+|targetj+) P(topic2-|targetj+)
P(iPhone+ |email+)
P(iPhone- |email+)
P(BB+ |email+)
P(BB- |email+)
![Page 102: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/102.jpg)
103
METHODOLOGY: LEARNING ASSOCIATIONS
Web search engine
Debate title
Topic1 = iPhoneTopic2 = BB
Weblogs containing both topics Pa
rser
Parsed web documents
Opinion-target pairing
Lexicon
Syntactic Rules
I like email = email+
Associations with
topic-polarity
P(iPhone- |email+)
P(BB- |email+)
P(iPhone+ |email+)
P(BB+ |email+)
like = +hate = -
![Page 103: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/103.jpg)
104
Blackberry+
Blackberry-iPhone-
iPhone+
Keyboard+
0.72
0.0
0.16
0.12
Associations learnt from web data
![Page 104: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/104.jpg)
105
Blackberry+
Blackberry-iPhone-
iPhone+
Keyboard- 0.25
0.25
0.125
0.375
Associations learnt from web data
0.5 0.5
![Page 105: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/105.jpg)
106
FROM THE WEB MINING PHASE
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Target-1 + Target-2 + Target-3 -
post
![Page 106: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/106.jpg)
107
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Target-1 + Target-2 + Target-3 -
post
? ?
![Page 107: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/107.jpg)
108
Stance-1Pro-iPhone
Stance-1Pro-Blackberry
iPhone + Blackberry - iPhone -Blackberry +
Target-1 + Target-2 + Target-3 -
post
Assume reinforcement unless detected otherwise
![Page 108: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/108.jpg)
109
Topic1+
Topic1-Topic2-
Topic2+
target+
Association of positive opinion towards a target to positive or negative opinions regarding either of the topics
Association Lookup
0.1
0.05
0.5
0.35
![Page 109: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/109.jpg)
110
Side-1 Side-2
Topic1+
Topic1-Topic2-
Topic2+
target+
Side-1 = Topic1+ alternatively Topic2-Side-2 =Topic2+ alternatively Topic1-
Association Lookup, Side Mapping
0.1
0.05
0.5
0.35
![Page 110: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/110.jpg)
111
target+Side-1 Side-2
0.150.85
Association of positive opinion towards a target to both of the stances
Association Lookup, Side Mapping
![Page 111: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/111.jpg)
112
NON-REINFORCING OPINIONS WITHIN THE POST
While the iPhone looks nice and does play a decent amount of music, it can't compare in functionality to the BB.
Concessionary opinions
Side Classification: pro-Blackberry
stance
![Page 112: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/112.jpg)
113
CONCESSION HANDLING
Detecting concessionary opinions Find Concession indicators
Discourse connectives from Penn Discourse Treebank (Prasad et al., 2007)
Use simple rules to find the conceded part of the sentence While the iPhone looks nice and does play a decent
amount of music, it can't compare in functionality to the BB.
I like my music, and phone, but I don't want to carry a brick around in my pocket when I only need my phone.
![Page 113: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/113.jpg)
114
Side-2Pro-Iphone
Side-1Pro-Blackberry
music+
phone+
1.0
0.5090.45
Original associations learnt from the web
CONCESSION HANDLING
![Page 114: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/114.jpg)
115
Side-2Pro-Iphone
Side-1Pro-Blackberry
music+
phone+
1.0
0.509 0.45
Associations after concession handling
Conceded opinions are counted for the opposite side
CONCESSION HANDLING
![Page 115: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/115.jpg)
116
Side-2Pro-Iphone
Side-1Pro-
Blackberry
Aggregation
target1+
target2+
target3+
target4+
Each opinion-target pair in the post has a bias toward one or the side
0.9
0.7
0.4
0.5
0.1
0.3
0.6
0.5
![Page 116: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/116.jpg)
117
Side-2Pro-Iphone
Side-1Pro-
Blackberry
Aggregation
target1+
target2+
target3+
target4+
Each opinion-target pair in the post has a bias toward one or the other sideAssign the side to the post which maximizes the association value of the post
![Page 117: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/117.jpg)
118
OTHER WORK
Multilingual Subjectivity (with Carmen Banea and Rada Mihalcea)
Opinion Implicatures (with Claire Cardie and Yejin Choi) Attitudes inferred from the explicit subjective
expressions in text Extract, aggregate and compare argument
expressions from multiple documents about controversial topics (with Alex Conrad and Rebecca Hwa) Annotation scheme for
Stance structures Argument expressions
![Page 118: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/118.jpg)
119
STANCE STRUCTURE
← sides
← aspects
← arguments
← debate root (obamacare)
![Page 119: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/119.jpg)
120
STANCE STRUCTUREAspects shared across sides
← pro arguments
←anti arguments
![Page 120: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/120.jpg)
121
ARGUING SPANS
“ObamaCare not only limits doctor-patient choice, it will -- if not reversed -- eventually force private insurance companies out of business and put everyone under a government-run system.”
• side: anti• arguing-against (alternative: “Obamacare”)• labels: hurts_private_insurance,
restricts_healthcare_choice
Arguing span:
![Page 121: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/121.jpg)
122
ARGUING SPANS
“Reform will finally bring skyrocketing health care costs under control, which will mean real savings for families, businesses and our government.”
• side: pro• arguing-for• label: controls_healthcare_costs
“We'll cut hundreds of billions of dollars in waste and inefficiency in federal health programs like Medicare and Medicaid and in unwarranted subsidies to insurance companies that do nothing to improve care and everything to improve their profits. ”
Arguing span:
Relevant supporting span:
![Page 122: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/122.jpg)
123
MANY OPEN PROBLEMS IN SUBJECTIVITY ANALYSIS
Complex discourse structure and pragmatics Non-literal language Irony and sarcasm Inferences and world knowledge
![Page 123: S UBJECTIVITY AND S ENTIMENT A NALYSIS : FROM W ORDS TO D ISCOURSE Jan Wiebe Department of Computer Science Intelligent Systems Program University of Pittsburgh](https://reader037.vdocuments.net/reader037/viewer/2022110206/56649d135503460f949e666e/html5/thumbnails/123.jpg)
124
THANK YOU