Building Efficient Mobile App Recommendation Systems Using Context-Aware Linear Regression Models

Abstract

The recommendation systems are imperative when it comes to the personal user experience in the mobile apps. But the traditional methods tend to perform poorly when it comes to taking into consideration the background information hence the poor performance. This paper seeks to have powerful recommendations of context sensible applications of cell phones with the help of convenient lineal models. The research is based on the fact that the gap in the literatures on the integration of the situational characteristic time of the day, user location, and user behavior history in using the app has to be closed. The analysis of the literature material proves that the deep learning models significantly increased the number of studies on the personalized recommendations (Smith et al., 2018), but the simplest, e.g., the linear regression, can produce the same good results, however, only when the contextual information will be introduced (Lee & Kim, 2020). The need to give the increased precision of the app suggestions on the basis of the use of the contextual data i.e. combination of some of its features in a linear regressor model stimulated the evolution of the suggested model. Unlike the traditional models, our context-aware model has registered improvement in its sound output both on a prediction and user interaction basis as it has been observed in our findings. The results can be used to develop systems of mobile application recommendations, which can show how the incorporation of the context into a linear regressive model can present the right and manageable alternative to the more elaborate designs. Future of research which is depicted through the proposed research is enhancement of the scalability of the model and combination of the real-time of the context and the recommendations.