Predicting Age Range of Users over Microblog Dataset

Abstract

In this paper, we present the idea and methodologies on predicting the age span of users over microblog dataset. Given a user's personal information such as user tags, job, education, self-description, and gender, as well as the content of his/her microblogs, we automatically classify the user's age into one of four predefined ranges. Particularly, we extract a set of features from the given information about the user, and employ a statistic-based framework to solve this problem. The measurement shows that our proposed method incorporating selected features has an accuracy of around 71% on average over the training dataset.