![Page 1: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/1.jpg)
M A N D Y K O R P U S I KP R O F E S S O R J U L I A H I R S C H B E R G
C O L U M B I A U N I V E R S I T YD E P A R T M E N T O F C O M P U T E R S C I E N C E
D I S T R I B U T E D R E U
A U G U S T 1 , 2 0 1 2
Tweet-to-Scene Generation: WordsEye with Twitter
![Page 2: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/2.jpg)
Outline
Introduction Background Method Results Future work
![Page 3: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/3.jpg)
WordsEye
Text-to-Scene Generation
Text: the orange cat is on the chair. the chair is white. the ground has a design texture. the ground is shiny. the huge yellow illuminator is 7 feet above the cat.
![Page 4: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/4.jpg)
Importance of Twitter
Over 100 million users Instantaneous communication Increasing popularity of social media Openly expressed opinions
![Page 5: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/5.jpg)
Twitter’s Language
140 characters limit Twitter lingo Hashtags (i.e. #justinbieber) Usernames (i.e. @msmith)
Internet lingo Acronyms (i.e. lol, omg) Emoticons (i.e. :D )
Misspellings Slang and swearing
![Page 6: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/6.jpg)
My Project
Tweet: “Yay!” Output: “The person is happy.”
![Page 7: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/7.jpg)
Getting Tweets
Random Filtered for a keyword Photo
![Page 8: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/8.jpg)
Normalizing Tweets
Replace emoticons with emotions Replace acronyms with words Extract hashtags and @usernames Replace words with repeated letters (i.e.
wooooooooooow)
![Page 9: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/9.jpg)
Filtering Tweets
Verbs Emotions
Like/Love Hate
![Page 10: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/10.jpg)
Extracting Topics
Tweet: “I love my white dog.”
Subject Object attribute
ObjectVerb
![Page 11: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/11.jpg)
FaceGen
![Page 12: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/12.jpg)
Active vs. Passive Verbs
Active “The woman likes the
flowers.” Subject -> Verb -> Object
Passive “The flowers are liked by
the woman.” Object -> Verb -> Subject
![Page 13: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/13.jpg)
Direct vs. Implied Emotions
Direct Emotion “I am so happy.” “He is bored.”
Implied Emotion “This movie is dull.” “The view is beautiful!”
![Page 14: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/14.jpg)
Emotional Verbs
Active “She loves music.” Subject -> Verb -> Object
Passive “That sound annoys me.” Object -> Verb -> Subject
![Page 15: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/15.jpg)
Text Generation
Verbs: “The [subject attribute] [subject] [verb] the [object attribute] [object].”
Emotions: “The [subject attribute] [subject] is [emotion] [preposition] the [object attribute] [object].”
Photos: “There is a small [photo filename] billboard.”
![Page 16: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/16.jpg)
Example Tweet
Normalized: “It’s a nice chill afternoon. Happy with the races today.”
Emotion words: “nice”, “happy” Output: “The person is happy about the chill
afternoon.”
![Page 17: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/17.jpg)
Results
“The woman hates the platypus.”
![Page 18: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/18.jpg)
Results
“The girl is excited about Justin Bieber.”
![Page 19: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/19.jpg)
Results
“There is a small [pic/1.jpg] billboard.”
![Page 20: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/20.jpg)
Results
“Barack Obama is angry at the Republican elephant.”
![Page 21: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/21.jpg)
Twitpic
Uploading WordsEyeimages
Traffic analysis
![Page 22: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/22.jpg)
Twitpic
![Page 23: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/23.jpg)
Tweeting Back
![Page 24: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/24.jpg)
Future Work
Include locations Add other WordsEye verbs Improve the topic extraction
![Page 25: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/25.jpg)
Questions?
![Page 26: Tweet-to-Scene Generation: WordsEye with Twitter · Twitter’s Language ... “The flowers are liked by ... Target-dependent Twitter Sentiment Classification. Proceedings of the](https://reader034.vdocuments.us/reader034/viewer/2022043018/5f3a94615c2c7b0fac0604c9/html5/thumbnails/26.jpg)
References
Das, D., & Bandyopadhyay, S. (2010). Extracting Emotion Topics from Blog Sentences -- Use of Voting from Multi-Engine Supervised Classifiers. SMUC, 119-126.
Esuli, A., & Sebastiani, F. (n.d.). SentiWordNet: A Publicly Available Lexical Resource for Opinion Mining. 117-122.
Gimpel, K., Schneider, N., O'Connor, B., Das, D., Mills, D., Eisenstein, J., et al. (2011). Part-of-Speech Tagging for Twitter: Annotation, Features, and Experiments. Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics, 42-47.
Gonzalez-Ibanez, R., Muresan, S., & Wacholder, N. (2011). Identifying Sarcasm in Twitter: A Closer Look. Proceedings of the 49th Annual meeting of the Association for Computational Linguistics, 581-586.
Han, B., & Baldwin, T. (2011). Lexical Normalisation of Short Text Messages: Makn Sens a #twitter. Proceedings of the 49th Annual meeting of the Association for Computational Linguistics, 368-378.
Jiang, L., Zhou, M., & Zhao, T. (2011). Target-dependent Twitter Sentiment Classification. Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics, 151-160.
Kim, S.-M., & Hovy, E. (2006). Extracting Opinions, Opinion Holders, and Topics Expressed in Online News Media Text. Proceedings of the Workshop on Sentiment and Subjectivity in Text, 1-8.
Liu, F., Weng, F., Wang, B., & Liu, Y. (2011). Insertion, Deletion, or Substitution? Normalizing Text Messages without Pre-categorization nor Supervision. Proceedings of the 49th Annual Meeting of the Association for Computational Linguistis, 71-76.
Qu, Z., & Liu, Y. (2011). Interactive Group Suggesting for Twitter. Proceedings of the 49th Annual meeting of the Association for Computational Linguistics, 519-523.
Stoyanov, V., & Cardie, C. (n.d.). Annotating Topics of Opinions. Zhao, W. X., Jiang, J., He, J., Song, Y., Achananupapr, P., Lim, E.-P., et al. (2011). Topical Keyphrase
Extraction from Twitter. Proceedings of the 49th Annual meeting of the Association for Computational Linguistics, 379-388.