Author_Institution :
Dept. of Comput., Sci. & Technol., Tsinghua Univ., Beijing, China
Abstract :
As the number of mobile applications grows rapidly, personalized recommendation for mobile apps becomes more and more important to mobile App stores. While most of the current works focused on proposing novel recommendation models for mobile apps, the lack of understanding of user download behaviors still remains a problem. Towards this end, we conducted a user behavior study on a large scale real world dataset, focusing on three kinds of biases to user download behaviors, namely, browsing history, app update history, and app categories. The dataset we used in this paper was constructed from Mobile Market (MM), an online mobile app store released by China Mobile, server logs from April 16, 2013 to June 13, 2013, containing 12,125,702 users and 67,577 apps.
Keywords :
mobile computing; recommender systems; app category; app update history; browsing history; large scale real world dataset; mobile market; online mobile app store; personalized recommendation; top-N recommendation model; user download behaviors; Computational modeling; Entropy; Games; History; Mobile communication; Recommender systems; Servers;
