Sentiment Analysis using Naive Bayes Classifier and Information Gain Feature Selection over Twitter

摘要

The development of the internet today is growing very rapidly which indirectly encourages the creation of personal web content that involves sentiments such as blogs, tweets, web forums and other types of social media. Humans often make decisions based on input from friends, relatives, colleagues and others. Supported by the availability of growth and popularity of opinionrich resources or sentiments such as online site reviews for ecommerce products and personal blogs For example, the expression of personal feelings that allows users to discuss everyday problems, exchange political views, evaluate services and products like Smartphone’s Smart TV’s etc. This research applies opinion mining method by using Na?ve Bayes Classifier and Information Gain algorithm based on Feature Selection. Testing this method uses the ECommerce based tweet dataset downloaded from the Twitter Cloud Repository. The purpose of this study is to improve the accuracy of the Na?ve Bayes algorithm in classifying documents along with Information Gain methodology. Accuracy achieved in this study amounted to 88.80% which is appropriate to evaluate the sentiments.