ABSTRACT
Sentiment analysis is a computational technique for studying attitudes, opinions, and emotions toward an object, such as a public figure or celebrity, events, or a trending topic, and the objects are either reviewed or rated. Sentimental mining uses natural language processing, data mining, and statistical techniques to identify, extract, and analyze opinions or sentiments of authors or speakers expressed in text or speeches. The techniques analyze and categorize sentiments as either positive or negative, although some classifications may include neutral sentiments. Of the many sentimental analysis techniques, this study evaluated and compared the performance of naive Bayes and support vector machine algorithms in classification sentiments embedded in movie reviews. In particular, the paper extends the polarity classification problem to MovieLens data; a big dataset that is 200 MB with over 20 million movie reviews. Unlike most articles that address the application of sentimental analysis in social media platforms, this study used the MovieLens review platform to extend the application of sentimental analysis techniques. The results of the study suggest that the support vector machine has a classification accuracy of 61%, a value lower than those established in other studies. However, the study established that support vector machine was most sensitive, since it had a recall value of 88.73%. Besides the positive and negative classes, the models explored very negative, very positive, somewhat negative, and somewhat positive, and this was the limitation of the study.
