Adi: Towards A Framework Of App Developer Inspection

With the popularity of smart mobile devices, the amount of mobile applications (or simply called apps) has been increasing dramatically in recent years. However, due to low threshold to enter app industry, app developers vary significantly with respect to their expertise and reputation in the production of apps. Currently, there is no well-recognized objective and effective means to profile app developers. As the mobile market grows, it already gives rise to the problem of finding appropriate apps from the user point of view. In this paper, we propose a framework called App Developer Inspector (ADI), which aims to effectively profile app developers in aspects of their expertise and reputation in developing apps. ADI is essentially founded on two underlying models: the App Developer Expertise (ADE) model and the App Developer Reputation (ADR) model. In a nutshell, ADE is a generative model that derives the latent expertise for each developer and ADR is a model that exploits multiple features to evaluate app developers' reputation. Using the app developer profiles generated in ADI, we study two new applications which respectively facilitate app search and app development outsourcing. We conduct extensive experiments on a large real world dataset to evaluate the performance of ADI. The results of experiments demonstrate the effectiveness of ADI in profiling app developers as well as its boosting impact on the new applications.