Lexicon-Based and Immune System Based Learning Methods in Twitter Sentiment Analysis

Nowadays, there are increasingly numbers of studies on seeking ways to mine Twitter for sentiment analysis. Machine learning approach such as immune system based learning methods is an alternative way for sentiment classification. This method is centered on prominent immunological theory as computation mechanisms that emulate processes in biological immune system in achieving higher probability for pattern recognition. The aim of this article attempts to study the potential of this method in text classification for sentiment analysis. This study consists of three phases; data preparation; classification model development using three selected Immune System based algorithms i.e. Negative Selection algorithm (NSA), Clonal Selection algorithm (CSA) and Immune Network algorithm (INA); and model analysis. As a result, NSA algorithm proposed slightly high accuracy in experimental phase and that would be considered as the potential classifiers for Twitter sentiment analysis. In future work, the accuracy of proposed model can be strengthened by comparative study with other heuristic based searching algorithms such as genetic algorithm, ant colony optimization, swam algorithms and etc.