SAIL: Sentiment Analysis using Semantic Similarity and Contrast Features

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 512–516, Dublin, Ireland, August 23-24, 2014. SAIL: Sentiment Analysis using Semantic Similarity and Contrast Features Nikolaos Malandrakis, Michael Falcone, Colin Vaz, Jesse Bisogni, Alexandros Potamianos, Shrikanth Narayanan Signal Analysis and Interpretation Laboratory (SAIL), USC, Los Angeles, CA 90089, USA {malandra,mfalcone,cvaz,jbisogni}@usc.edu, potam@telecom.tuc.gr, shri@sipi.usc.edu Abstract This paper describes our submission to Se- mEval2014 Task 9: Sentiment Analysis in Twitter. Our model is primarily a lexi- con based one, augmented by some pre- processing, including detection of Multi- Word Expressions, negation propagation and hashtag expansion and by the use of pairwise semantic similarity at the tweet level. Feature extraction is repeated for sub-strings and contrasting sub-string fea- tures are used to better capture complex phenomena like sarcasm. The resulting supervised system, using a Naive Bayes model, achieved high performance in clas- sifying entire tweets, ranking 7th on the main set and 2nd when applied to sarcastic tweets. 1 Introduction The analysis of the emotional content of text is relevant to numerous natural language process- ing (NLP), web and multi-modal dialogue appli- cations. In recent years the increased popularity of social media and increased availability of rele- vant data has led to a focus of scientific efforts on the emotion expressed through social media, with Twitter being the most common subject. Sentiment analysis in Twitter is usually per- formed by combining techniques used for related tasks, like word-level (Esuli and Sebastiani, 2006; Strapparava and Valitutti, 2004) and sentence- level (Turney and Littman, 2002; Turney and Littman, 2003) emotion extraction. Twitter how- ever does present specific challenges: the breadth of possible content is virtually unlimited, the writ- ing style is informal, the use of orthography and This work is licensed under a Creative Commons At- tribution 4.0 International Licence. Page numbers and pro- ceedings footer are added by the organisers. Licence details: http://creativecommons.org/licenses/by/4.0/ grammar can be “unconventional” and there are unique artifacts like hashtags. Computation sys- tems, like those submitted to SemEval 2013 task 2 (Nakov et al., 2013) mostly use bag-of-words models with specific features added to model emo- tion indicators like hashtags and emoticons (Davi- dov et al., 2010). This paper describes our submissions to Se- mEval 2014 task 9 (Rosenthal et al., 2014), which deals with sentiment analysis in twitter. The sys- tem is an expansion of our submission to the same task in 2013 (Malandrakis et al., 2013a), which used only token rating statistics as features. We expanded the system by using multiple lexica and more statistics, added steps to the pre-processing stage (including negation and multi-word expres- sion handling), incorporated pairwise tweet-level semantic similarities as features and finally per- formed feature extraction on substrings and used the partial features as indicators of irony, sarcasm or humor. 2 Model Description 2.1 Preprocessing POS-tagging / Tokenization was performed using the ARK NLP tweeter tagger (Owoputi et al., 2013), a Twitter-specific tagger. Negations were detected using the list from Christopher Potts’ tutorial. All tokens up to the next punctuation were marked …

Content truncated to 500 words