Proposed Approach for Sarcasm Detection in Twitter

Background: Sarcasm detection in twitter is a very important task as it had helped in the analysis of tweets. With the help of sarcasm detection, companies could analyze the feelings of user about their products. This is helpful for companies, as the companies could improve their quality of product. Methods: For preprocessing of data TextBlob is used. TextBlob is a package installed in Natural Language Toolkit. The preprocessing steps include tokenization, part of speech tagging and parsing. The stop words are removed using python programming. The stop words corpus which consist of 2400 stop words and which is distributed with NLTK have been used. RapidMiner is used for finding polarity and subjectivity of tweets. TextBlob is used for finding the polarity and subjectivity confidence. Weka is used for calculating the accuracy of tweets based on Naïve Bayes classifier and SVM classifiers. Findings: The paper provides the polarity of tweets which include whether the tweet is positive, negative or neutral. Polarity confidence and subjectivity confidence are also found. Accuracy of tweets are found using Naïve Bayes and SVM classifiers. Applications: Sarcasm Detection could be helpful in analyzing the exact opinion of the user about a certain thing.