Recognition of Sarcasms in Tweets Based on Concept Level Sentiment Analysis and Supervised Learning Approaches

PACLIC 28 !404 Recognition of Sarcasm in Tweets Based on Concept Level Sentiment Analysis and Supervised Learning Approaches Piyoros Tungthamthiti, Kiyoaki Shirai, Masnizah Mohd Japan Advanced Institute of Science and Technology 1-1, Asahidai, Nomi City, Ishikawa, Japan 923-1292 Email: {s1320204,kshirai,masnizah}@jaist.ac.jp Abstract Sarcasm is a form of communication that is intended to mock or harass someone by us- ing words with the opposite of their literal meaning. However, identification of sarcasm is somewhat difficult due to the gap between its literal and intended meaning. Recognition of sarcasm is a task that can potentially pro- vide a lot of benefits to other areas of nat- ural language processing. In this research, we propose a new method to identify sarcasm in tweets that focuses on several approaches: 1) sentiment analysis, 2) concept level and common-sense knowledge 3) coherence and 4) machine learning classification. We will use support vector machine (SVM) to classify sarcastic tweet based on our proposed features as well as ordinary N-grams. Our proposed classifier is an ensemble of two SVMs with two different feature sets. The results of the experiment show our method outperforms the baseline method and achieves 80% accuracy. 1 Introduction Recognition of sarcasm is one of the most difficult tasks in natural language processing (NLP). It is a problem of determining if the actual meaning of a word is intended in a given context. Sarcasm is normally represented in a form of ironic speech in which the speakers convey an implicit message to criticize a particular person. Thus, tone of voice plays a significant role in the communication. There are many communication programs (e.g. Line, Face- book, Twitter), which allow to communicate to- gether through only text characters. It is very diffi- cult to determine the actual meaning by just looking at the text itself. Recognition of sarcasm prevents us from misinterpreting sentences whose meaning are opposite to their literal meaning. It is also a task that is potentially applicable for many other areas of NLP, for example, machine translation, information retrieval, information extraction and knowledge ac- quisition. Twitter is an online social networking service that allows users to post and read short messages, called “tweets”. However, Twitter allows users to write short messages, i.e. 140 characters per tweet. Also, users usually post a lot of tweets in complex sen- tence structures. Regarding to these issues, a new method is created to detect sarcasm in tweets. Sarcasm is known as “the activity of saying or writing the opposite of what you mean, or of speak- ing in a way intended to make someone else feel stupid or show them that you are angry” (Macmil- lan, 2007). According to this definition, we can rec- ognize sarcasm by evaluating the polarity of the sen- tences. In other words, a sarcastic sentence contains two or more words, which may cause conflict in sen- timent polarities (both positive and negative) in a sentence, whereas a normal sentence should contain at most one polarity. Let us consider the example sentence “I love …

Content truncated to 500 words