Development of a Distributed Workflow-based Analysis Service for Metadata of Mobile Applications

Universität Ulm

MA Abschlussvortrag, Jörn Hofschlaeger, Ort: Online, Datum: 15.07.2020, Zeit: 11:00 Uhr

Smartphones have multiple functions and can, therefore, be used for many different applications. An interesting use case is in the area of mHealth apps. Therapists can use mHealth apps for special treatments in the field of meditation, depression, or tinnitus, to name just a few. This poses a problem because the ratings of apps are not objective and can be misleading. This leads to the difficulty that therapists are not able to make a good decision based on the information provided.
To improve this situation, a workflow-based service for the use case of the analysis of metadata for mHealth apps is developed. This service retrieves the metadata of the apps from the Google Play Store and allows to save the metadata of different points in time. In addition, experts can rate the apps using the MARS questionnaire in order to extend the existing data. The advantage of this service is that more information about mHealth apps is available, and the evaluations are more objective.
By using the metadata of an app at different points in time, changes can be analyzed. The analysis of app metadata and the questionnaire data is used for the creation of a user interface that provides an overview of the changes and shows the results of the questionnaires. Consequently, this will help therapists to determine whether the app is suitable for a specific case.
In this thesis, a workflow engine for the orchestration of microservices is used. This is a modern approach to achieve a maintainable and scalable solution. The core of the presented solution is based on Zeebe, a product for orchestration of service tasks defined in a BPMN 2.0 workflow. Each service task is implemented as a microservice. The service tasks are implemented using the Zeebe client and are developed in the modern programming language Go. To store the data, CouchDB is used. An open-source web scraper written in JavaScript is used to retrieve the app metadata from the Google Play Store. The prototype presented in this thesis shows that a service for analyzing metadata of mobile applications can be based on the technologies used.