Influence Based Group Recommendation System in Personality and Dynamic Trust

Given the frequent engagement in group activities within daily life, recommending content to a group of users becomes an important task. In heterogeneous groups, a conflict situation may arise more easily if the preferences of group members are incompatible. The challenge with dynamic groups lies in reconciling the diverse preferences of its members to reach a collective decision that satisfies everyone. While social dynamics such as personality traits, and mutual influence play a pivotal role in shaping group decision-making, this study employs the TKI personality traits, which have demonstrated efficacy in mitigating conflicts during group decision processes. Besides, we have developed a novel dynamic trust mechanism that adeptly captures the evolving trust values within a group integrated into our refined group recommendation algorithms. In order to achieve our research objectives, we executed a four-week empirical study by deploying a responsive web application tailored for our group recommendation system. Users in the experiment interacted with two distinct algorithms: traditional influence-based aggregation and the influence matrix algorithm, each over the course of two consecutive weeks. Through the experiment, we are able to better capture the variability of social factors in group decision-making, achieving higher accuracy and satisfaction as well as laying the foundation for a milestone in group recommendation within the restaurant domain.