A Novel APPs Recommendation Algorithm Based on APPs Popularity and User Behaviors

With the proliferation of smart phones, mobile applications (APPs) are increasingly being used for mobile work and entertainment. In order to satisfy people's various demands, there are enormous mobile applications being delivered through different mobile application markets. This brings markets owners huge opportunities and tough challenges simultaneously. It is very difficult to find proper APPs for users within such large number of APPs. To alleviate this problem, traditional recommendation techniques are introduced into APPs recommendation. However, different from traditional context, APPs recommendation is a very unique task since people use APPs for different reasons. In this paper, we analyzed user's usage and download behaviors based on a real Android Market data to seek useful information which can benefit APPs recommendation task. We present a new matrix factorization algorithm which incorporates APPs popularity and user behaviors. The experiment shows our method outperforms traditional recommendation approaches in mobile recommendation context.