Ecosystem Evolution Analysis and Trend Prediction of Projects in Android Application Framework

Abstract

The application framework layer in the Android system consists of numerous project repositories, which rely on each other to form a co-evolving software ecosystem. Android's application framework layer provides many useful APIs to millions of Android Apps, so its evolution will affect the robustness and stability of Android Apps. Code dependency analysis technology is a common way to analyze software ecosystems. However, the code size of projects in the Android application framework layer is so huge that ordinary analysis methods are unacceptable due to the excessive resources required. In this paper, we propose an approach for evolution analysis and trend prediction based on the subgraph of code dependency network graph, in order to realize the effective analysis of large-scale software ecosystem. Based on the source code data of the application framework collected from AOSP, our proposed approach is verified. The prediction results of our model show that the average values of precision and recall are 90.0% and 90.4% respectively, which proves that our approach can well is effective.