Sentiment analysis on Twitter using the combination of lexicon-based and support vector machine for assessing the performance of a television program

The development of social media, especially Twitter is growing rapidly. Twitter is usually used to comment on a product, a person or even a television program. The written comments by Twitter users can reach hundred thousand or even millions every day. By using the comments obtained from Twitter, it can complement a television program assessment that usually done by using rating, which only represented in terms of quantity. Therefore, by analyzing the comments on Twitter hopefully it would be able to complement the assessment in terms of quality. The comments from Twitter can be analyzed by performing a sentiment analysis process. The methods used in this study are the combination of lexicon-based method and Support Vector Machine. The results show that these combination methods can be implemented in analyzing sentiment on the television program with the accuracy rate that reaches 80%. This value is not influenced by the ratio of training data and test data that are used. However, in this study the data tweets that are dominant by positive sentiment tends to have a higher accuracy rate than the data tweet that consist a balanced amount of sentiment or the one that are dominant by the negative sentiment.