Age and Gender Identification by SMS Text Messages

Abstract

In this study, age and gender identification are tried to be predicted from SMS text messages. 38,588 preprocessed text messages were tested which were written by native English and Singaporean English students. Naive Bayes, Support Vector Machine, and J48 Decision tree are applied for gender identification and age range prediction of the author of a given text messages. The test resulted in 70.79% average accuracy for correct age prediction with Support Vector Machine algorithm, and 79.10% average accuracy for correct gender identification via using J48 decision tree.