Using Feedback to Promote Meaningful App-Switching Suggestions

In the last years, smartphones obtained a role of growing relevance and presence in our lives, proving to be not only a communication tool, but also a tool for accessing various functionalities in an easy way. For example, users may read their emails and immediately insert and event on Calendar, while also chatting with a friend on Telegram. Among all the recent studies that analyzed smartphone usage and its consequences under different aspects, one emerging and interesting field of research regards the so-called app switching behavior, i.e., “transitioning from one app to another in the same usage session to consume content”. An efficient strategy to automatically extracting and characterizing app switching behaviors and support these switches in modern smartphone was implemented in RecApps (Alberto Monge Roffarello, Luigi De Russis), a mobile application that analyzes the user's smartphone usage, collecting it into a phone sessions representation, and then extract association rules representing habitual switching behaviors, using them as links to other apps. This work, however, tends to not take into consideration information about the user, possibly disregarding their digital wellbeing, i.e., “a state where individual comfort is preserved despite an environment characterized by the overabundance of digital communication”. This means that potentially negative behaviors and habits may be reinforced and proposed to users. The goal of this Thesis is to design a new recommending system, taking RecApps as a starting point, and implement strategies to consider users’ digital wellbeing in the process of computing and promoting app-switches. Prior to the actual implementation phase, several materials were studied to understand the fundamental concepts of digital wellbeing and how it can be supported through technological means (i.e., the so-called “Digital Self Control Tools”), then a design phase was dedicated to considering how to integrate digital wellbeing elements into the suggestion architecture of RecApps. Given the objective nature of the existing collected information, an approach based on proactive feedback given by the user, and attached to each phone session, was considered, in order to guide the transition extraction procedure and include also some contribution of the user in the entire analysis. The chosen strategy was implemented by adding a form for collecting feedback information from users after a phone session has ended, and then consider these new elements into the rule extraction phase and recommendations suggestion. After the design and implementation phases, an in-the-wild user study was conducted to assess whether the adopted strategy and the resulting application were effective in supporting meaningful interactions and switching behaviors. Results were obtained through the use of objective metrics on the collected data, and also the analysis of the final interviews to the participants. The obtained results shown that adding user feedback to the collected information and include these additional inputs as part of the recommendation mechanism promoted transitions that effectively reflected such feedbacks, highlighting transitions that were considered meaningful and lowering the frequency of others that were rated negatively.