Date of Award
Doctor of Philosophy (PhD)
Modern software platforms feature digital distribution channels called marketplaces, which have revolutionized the way applications are developed and delivered to users. As the number of applications continues to proliferate in marketplaces, the need to fully understand them is ever increasing. While researchers have recently started to observe the wealth of information in marketplaces, their efforts have been largely constrained to one view of analysis and a single snapshot in time. As a result, the increasing number of application updates published to marketplaces has largely gone unobserved. Such view misses the much larger opportunity of mining applications with both a deep and longitudinal views and utilizing it to create innovative systems.
This dissertation introduces a new approach to analyzing large, ever-evolving marketplaces, such as the official Android marketplace, by taking a deep and longitudinal perspectives. To make this approach feasible, I designed and developed a scalable platform called Sieveable. Sieveable provides efficient retrieval of hundreds of thousands of applications with the goal of enabling a deep and longitudinal analysis of the design and development of mobile applications.
I demonstrate how Sieveable enables different types of analyses in three main areas that would have been very difficult to perform otherwise. In user interface design, the release of official design libraries enabled new applications to narrow the gap with the most downloaded ones in adopting best design practices. In accessibility, results showed that accessibility is a problem for many applications including the most downloaded ones. In privacy, the most added permissions in each year are the ones often required by ad libraries, which raises privacy concerns. The findings of this work offer insights to marketplace owners, platform engineers, and developers. I argue that considering both a deep and a longitudinal views results in a more useful analysis to support the design and development of mobile applications.
Alharbi, Khalid Ahmed, "A Deep and Longitudinal Approach to Mining Mobile Applications" (2016). Computer Science Graduate Theses & Dissertations. 152.