Recognizing Stances in Online Debates 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 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology.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 … ACL 2009

Recognizing Stances in Online Debates 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 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology.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 … ACL 2009 Side Classification: pro-iPhone stance

Recognizing Stances in Online Debates 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 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology.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 … ACL 2009 Arguing why their stance is correct

Recognizing Stances in Online Debates 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 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology.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 … ACL 2009 Justifying why the opposite side is not good

ACL

ACL Side Classification: pro-iPhone stance Side Classification: pro- Blackberry stance Side Classification: pro-iPhone stance Topics: 1.iPhone 2.Blackberry Sides/ Stances: 1.Pro-iPhone 2.Pro-Blackberry Dual-topic, Dual-sided debates regarding Named Entities

Goal Debate stance recognition using opinion analysis Learn debating preferences from the web Exploited in an unsupervised approach Combines the individual pieces of information to classify the overall stance ACL 2009

Challenges ACL 2009 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology

Challenges ACL 2009 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology Positive and negative opinions are employed to argue for a side Side Classification: pro-iPhone stance

Challenges ACL 2009 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology Positive and negative opinions are employed to argue for a side Opinions towards both topics within a post Side Classification: pro-iPhone stance

Challenges ACL 2009 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology Positive and negative opinions are employed to argue for a side Opinions towards both topics within a post Side Classification: pro-iPhone stance + towards iPhone - towards Blackberry

Challenges ACL 2009 The iPhone incarnate the 21st century whereas Blackberry symbolizes an outdated technology Positive and negative opinions are employed to argue for a side Opinions towards both topics within a post Side Classification: pro-iPhone stance + towards iPhone - towards Blackberry We need to consider not only positive and negative opinions but also what they are about (targets)

Challenges 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. ACL 2009

Challenges 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. Debate topics are evoked in a variety of ways: Opinions explicitly toward the named topics are not enough Type of Blackberry Feature of Blackberry Maker of iPhone Feature of iPhone ACL 2009

Challenges 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. We need to consider not only opinions towards topics, But also opinions towards aspects ACL 2009

Challenges 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. We need to consider not only opinions towards topics, But also opinions towards aspects Unique Aspects ACL 2009

Challenges iPhone and Blackberry, both Have facilities Can be used to take photos Operate on batteries Etc. Both sides share aspects ACL 2009

Challenges … I love the ability to receive s from my work account… ACL 2009 People expressing positive opinions regarding s (generally) prefer Blackberry

Challenges … I love the ability to receive s from my work account… ACL 2009 Certain shared aspects may be perceived to be better in one side on Blackberry Value for shared aspects depends on personal preferences ing – pro-Blackberry people will argue via + That is, + is often a strategy for arguing for the pro-Blackberry stance. Or, Browsing+ for iPhone

Challenges … I love the ability to receive s from my work account… ACL 2009 We need to find what a preference/dislike for an individual target means towards the debate stance as a whole

Challenges While the iPhone looks nice and does play a decent amount of music, it can't compare in functionality to the BB. Concessionary opinions can be misleading ACL 2009 Side Classification: pro- Blackberry stance

Challenges While the iPhone looks nice and does play a decent amount of music, it can't compare in functionality to the BB. ACL 2009 Side Classification: pro- Blackberry stance We need to detect and handle concessionary opinions

Challenges: Summary For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions ACL 2009

Challenges: Summary For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions ACL 2009 Turney, 2002; Pang et al, 2002; Dave et al, 2003; Yu and Hatzivassiloglou, 2003, Pang and Lee 2005, Wilson et al 2005, Goldberg and Zhu, 2006, Kim and Hovy 2006 …

Challenges: Summary For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions ACL 2009 Hu and Liu, 2004; Popescu and Etzioni., 2005; Bloom et al. 2007, Stoyanov and Cardie 2008; Xu et al., 2008 …

For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions Our Approach ACL 2009

For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions Our Approach ACL 2009 Adopting from previous work, Opinion-target pairing using Opinion Lexicons and Syntactic rules

For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions Our Approach ACL 2009 Unsupervised system Learn Associations from web and incorporate these towards stance recognition Adopting from previous work, Opinion-target pairing using Opinion Lexicons and Syntactic rules

For debate stance recognition we need to: Consider not only positive and negative opinions, but also what they are about (targets). Consider not only opinions towards topics, but also opinions towards aspects Find what a preference/dislike for an individual target means towards the debate stance as a whole Detect and handle concessionary opinions Our Approach ACL 2009 Unsupervised system Learn Associations from web and incorporate these towards stance recognition Adopting from previous work, Opinion-target pairing using Opinion Lexicons and Syntactic rules Rule-based Concession Handler using PDTB connectives

Methodology Learn associations from web data (weblogs) Process the web data to Find opinion-target pairs Associate opinion-target pairs with each debate side Utilize the associations to classify debate posts Process the debate posts to Find opinion-target pairs in the post Handle concessionary opinions Optimize over all opinion-targets for a post-level stance classification ACL 2009

Methodology Learn associations from web data (weblogs) Process the web data to Find opinion-target pairs Associate opinion-target pairs with each debate side Utilize the associations to classify debate posts Process the debate posts to Find opinion-target pairs in the post Handle concessionary opinions Optimize over all opinion targets for a post-level stance classification ACL 2009

Methodology: Learning associations ACL 2009 Debate title Topic 1 = iPhone Topic 2 = BB

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Yahoo search engine API

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Yahoo search engine API

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Stanford parser

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Stanford parser

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing Lexicon like: + hate: - Wilson et al., 2005

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing Lexicon Syntactic Rules like = + hate = -

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing Lexicon Syntactic Rules I like = + like = + hate = -

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing Lexicon Syntactic Rules I like = + Associati ons with topic- polarity like = + hate = -

Topic1+ Topic1- Topic2- Topic2+ target j + what does a positive opinion towards a target mean with respect to positive or negative opinions regarding either of the topics Associations with topic-polarity ACL 2009

Associations with topic-polarity For each opinion-topic pair (topic 1 +, topic 1 -, topic 2 +, and topic 2 -) found in the web document Find other opinion target pairs (target j p ) in its vicinity For each opinion-target (target j p ) calculate its association with each of the opinion-topics P(topic 1 +|target j +) P(topic 1 -|target j +) P(topic 2 +|target j +) P(topic 2 -|target j +) ACL 2009 P(iPhone+ | +) P(iPhone- | +) P(BB+ | +) P(BB- | +)

Methodology: Learning associations ACL 2009 Web search engine Debate title Topic 1 = iPhone Topic 2 = BB Weblogs containing both topics Parser Parsed web documents Opinion- target pairing Lexicon Syntactic Rules I like = + Associati ons with topic- polarity P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +) like = + hate = -

Blackberry+ Blackberry- iPhone- iPhone+ Storm Associations with topic-polarity ACL 2009

Methodology Learn associations from web data (weblogs) Process the web data to Find opinion-target pairs Associate opinion-target pairs with each debate side Utilize the associations to classify debate posts Process the debate posts to Find opinion-target pairs in the post Handle concessionary opinions Optimize over all opinion targets for a post-level stance classification ACL 2009

Methodology: Stance Classification ACL 2009 Debate Post

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = +

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +)

Topic1+ Topic1- Topic2- Topic2+ target+ Association of positive opinion towards a target to positive or negative opinions regarding either of the topics Association Lookup ACL 2009

Side-1 Side-2 Topic1+ Topic1- Topic2- Topic2+ target+ Side-1 = Topic1+ alternatively Topic2- Side-2 =Topic2+ alternatively Topic1- Association Lookup, Side Mapping ACL 2009

target+ Side-1 Side Association of positive opinion towards a target to both of the stances Association Lookup, Side Mapping ACL 2009

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairingin the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +)

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairingin the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +) Assoc(Side-1, +) Assoc(Side-2, +)

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +) Concession Handling Assoc(Side-1, +) Assoc(Side-2, +)

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. ACL 2009

Concession Handling 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. Conceded opinions music+ phone+ ACL 2009

Side-2 Pro-Iphone Side-2 Pro-Iphone Side-1 Pro-Blackberry Side-1 Pro-Blackberry music+ phone Original associations learnt from the web Concession Handling ACL 2009

Side-2 Pro-Iphone Side-2 Pro-Iphone Side-1 Pro-Blackberry Side-1 Pro-Blackberry music+ phone Concession Handling Associations after concession handling Conceded opinions are counted for the opposite side ACL 2009

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +) Concession Handling Post-level association aggregation Post-level association aggregation Assoc(Side-1, +) Assoc(Side-2, +)

Side-2 Pro-Iphone Side-2 Pro-Iphone Side-1 Pro- Blackberry Side-1 Pro- Blackberry Aggregation target 1 + target 2 + target 3 + target 4 + Each opinion-target pair in the post has a bias toward one or the side ACL 2009

Side-2 Pro-Iphone Side-2 Pro-Iphone Side-1 Pro- Blackberry Side-1 Pro- Blackberry Aggregation target 1 + target 2 + target 3 + target 4 + Each opinion-target pair in the post has a bias toward one or the other side Optimize the post classification such that The side assigned to the post maximizes the association value of the post ACL 2009

Methodology: Stance Classification ACL 2009 Debate Post Parser Debate Post Parsed Debate Post Opinion- target pairing in the post Lexicon Syntactic Rules I like = + Association lookup, Side Mapping P(iPhone- | +) P(BB- | +) P(iPhone+ | +) P(BB+ | +) Concession Handling Post-level association aggregation Post-level association aggregation Side= pro-Topic1 Assoc(Side-1, +) Assoc(Side-2, +)

Blackberry+ Blackberry- iPhone- iPhone+ Storm Associations learnt from web data ACL 2009

Blackberry+ Blackberry- iPhone- iPhone+ Storm Associations learnt from web data Both OpPMI, and Op-Pref agree with each other; Both learnt the IS-A relationship ACL 2009

Blackberry+ Blackberry- iPhone- iPhone+ Keyboard Associations learnt from web data ACL 2009

Blackberry+ Blackberry- iPhone- iPhone+ Keyboard Associations learnt from web data Negative opinions towards keyboards are not useful for side discrimination 0.5 ACL 2009

Summing Up Looked at several tasks ranging from purely lexical to discourse classification Identify subjective words Classify their senses as subjective or objective Recognize, in a text or conversation, whether a word is used with a subjective or objective sense Sense-aware contextual subjectivity and sentiment analysis Contextual polarity recognition Discourse-Level Opinion Interpretation Many ambiguities are involved in interpreting subjective language!

Summing Up Many other ambiguities than these! Sarcasm and Irony Yeah, he’s just wonderful. You’re no different from the mob! Oh, there’s a big difference, Mrs. De Marco. The mob is run by murdering, thieving, lying, cheating psychopaths. We work for the President of the United States. [Married to the Mob] Literal versus non-literal language He is a pain in the neck

Pointers Please see Publications OpinionFinder Subjectivity lexicon MPQA manually annotated corpus Tutorials Bibliography

Acknowledgements Subjectivity Research Group, Pittsburgh Cem Akkaya, Yaw Gyamfi, Paul Hoffman, Josef Ruppenhofer, Swapna Somasundaran, Theresa Wilson Cornell: Claire Cardie, Eric Breck, Yejin Choi, Ves Stoyanov Utah: Ellen Riloff, Sidd Patwardhan, Bill Phillips UNT: Rada Mihalcea, Carmen Banea Wendy Chapman, Rebecca Hwa, Pam Jordan, Diane Litman, …

Thank you ACL 2009