Gender Prediction Based on Data Streams of Smartphone Applications

摘要: Gender information has great values in personalized service, targeted advertising, recommender systems and other aspects. However, such is kind of private information, that many users are reluctant to share. In this paper, we propose a novel approach predict the users’ gender by analyzing data streams smartphone applications. The proposed assumes certain features extracted from could represent perspective characteristics (e.g., gender). To be more specific, noticed male female have different response time Thus extract key feature – Response-Time application. Moreover, leveraging construct training data, further importing Support Vector Machine (SVM) classifier, verified well predicted. experiments, dataset real world collected 25 volunteers. prediction results can achieve 86.50% Accuracy 86.43% F1-score, respectively. best our knowledge, first was predicted