Hybrid Method for Sarcasm Target Identification to Assist the Sentiment Analysis Systems

Savita B. Borole

At present the online social networking sites like twitter is very common and becomes important part of human beings for different purposes. The sarcasm is sophisticated form of irony and frequently used by the end users on such social network applications. The sarcasm is mainly used to indicate implicit information inside the post a person transmits on social networks like mockery, criticism etc. The detection of such sarcasm is very difficult tasks manually; hence it is very important to recognize the sarcasm targets from sarcastic texts automatically. The extraction of sarcasm targets in messages helps to improve the systems like automatic sentiment analysis, opinion mining, human treats analysis etc. In this paper, the main focus to solve the problem of accuracy sarcasm target extraction from the sarcastic texts to boost the performance of sentiment analysis systems. The process of extracting the target of ridicule in a sarcastic sentence is known as sarcasm target extraction. The preliminary hybrid model designed based on the rule based identification and classifier based identification. We defined 12 rules with inclusion of emoticons, emotional, and hashtag based rules. The 70 % of users used the emoticons, emotional texts, and hastag to express their feelings in sarcastic texts. The statistical classifier additionally used to extract the sarcasm target from the text. The fusion of both methods delivers the set of sarcasm targets from the sarcastic texts. The experimental analysis using the sarcasm dataset proves the efficiency of proposed method.

Volume 11 | 04-Special Issue

Pages: 405-411