An Improved Feature Selection Method for Short Text Classification

Abstract

Text has become one of the widest means of communication on mobile devices due to cheap rate and convenience for instance short text, web document, emails, instant messages. The exponential growth of text documents shared among users globally has increased the threat of misclassification associated with mobile devices such as Spam, Phishing, License to kill, Malware and privacy issues. Existing studies have shown that the major problem associated with text message classification is the poor representation of feature thus reducing accuracy and increasing f-measure rate. Thus, a modified Genetic Algorithm (GA) for improve feature selection and Artificial Immune System (AIS) algorithm was proposed for effective text classification in mobile short messages. The system will be deployed on an Android OS.