Abstract
Traditional requirements elicitation methods typically involve interviews, observations, questionnaires, prototyping, etc. Despite their usefulness, these methods heavily depend on the knowledge of stakeholders and requirements engineers. In the fast-paced and highly competitive mobile app market, staying ahead of evolving trends is particularly challenging. App stores like Google Play and the Apple Store offer a vast repository of apps, providing an opportunity to identify similar products and gain valuable insights. However, the sheer volume of available apps makes manual analysis a daunting and time-consuming task. To bridge this gap, our research focuses on leveraging app store data to streamline and enhance the requirements elicitation process.In this thesis, we focus on three challenges of requirements elicitation: refinement of initial idea, rapid and accessible prototyping, and continuous requirements elicitation after the app's release. To tackle these challenges, we developed three innovative approaches that leverage app descriptions, introduction images, and app reviews from app stores. First, we proposed a method to identify relevant app descriptions and extract key features for sub-feature recommendation, which helps refine high-level ideas/features into detailed features. Second, by mining app introduction images from Google Play, we developed a GUI search engine that allows users to quickly find relevant screenshots based on textual queries, thereby accelerating the prototyping process. Third, we introduced an automated pipeline to extract requirements-related information from large volumes of app reviews. Empirical evaluations demonstrate the effectiveness of these three approaches. Additionally, we conducted a case study on the requirements elicitation process for a health monitoring app designed for seniors, which further validated their effectiveness in a real-world scenario.