Providing Assurances for Self-Adaptation in a Mobile Digital Storytelling Application Using ActivFORMS

Abstract

Self-adaptability enables a system to adapt itself to changes in its execution conditions and user requirements in order to achieve particular quality goals. However, assuring that the adaptation goals are satisfied poses complex challenges. We recently developed the ActivFORMS approach that aims to tackle some of these challenges, but further research is required to evaluate the approach. This paper presents the results of a study in which we applied ActivFORMS to a mobile storytelling application that employs a social recommender. The initial version of the application used a static recommender that could not deal with changing environment conditions, or take into account preferences of users. To that end, we added a self-adaptive layer on top of the application. The study results show that self-adaptation significantly increases the quality of recommendations compared to the initial version by: (1) enabling the social recommender to adapt to the quality of user input and unavailability of the GPS service, and (2) making the recommender adaptive to user preferences. Providing guarantees for these adaptation goals is crucial in this domain from a business perspective. The study results show the feasibility and effectiveness of ActivFORMS for a practical application, but they also underpin the need for an integrated verification approach for self-adaptive systems that combines offline with online verification.