Paper Title
Mapping Sentiment Analysis Research Evolution

Abstract
Sentiment classification is an integral part of sentiment analysis. The interest to classify non-structured text into positive and negative by narrowing the margin of obtaining a neutral score has encouraged researchers to look for new classification method that is not just accurate but is able to replicate the results using different corpus within the same domain. Machine learning algorithms trains a small amount of data for classification purposes and has the ability to predict future patterns in a shorter time span. This paper is interested in mapping out the current research within the sentiment analysis framework that solely adopts machine learning techniques for classification purpose. A total of 1,711 papers have been identified and mapped out based on systematic mapping guidelines. Results of the mapping process reveal most of the recent publications adopt machine learning approach in order to evaluate their proposed methods and frameworks by means of experimentation. The basis of this paper is to provide a wide overview of the research area. The result of this mapping study can identify areas suitable for conducting systematic literature reviews and also areas where a primary study is more suitable. Index Terms - Sentiment classification, sentiment analysis, research trends, systematic mapping.