Abstract
The exponential growth of information available on the Internet has created a pressing need for tools that assist users in making informed decisions. Recommender Systems address this challenge by suggesting items tailored to individual preferences. However, certain decision-making scenarios involve shared experiences among groups, such as dining, traveling, or entertainment, where individual RSs fail to meet the unique requirements of group decision-making. To fill this gap, Group Recommender Systems (GRS) have been developed, offering tools and techniques designed to support group decision-making by recommending items relevant to the collective preferences of a group and providing other types of support for group discussions. This thesis investigates the dynamics of group decision-making and the role of GRSs in facilitating effective outcomes of group decision-making. We explore decision-making processes in both for a group, where decisions are made on behalf of other group members, and within a group, where all members actively contribute. The research identifies gaps in existing systems. By addressing these gaps, we propose innovative supporting functionalities and provide deeper insight into group decision-making to enhance the GRS’s ability to support group discussions and lead them to satisfactory decisions. Specifically, we gained deeper insights into group decision-making by conducting a focus group study, the outcomes of which serve as guidelines for designing effective user interactions tailored to real user requirements. Next, we addressed the challenge of decision making on behalf of a group by developing a prototype to support individual users in making decisions on behalf of the group. Finally, we proposed a group choice prediction model that identifies a key dimension of the group’s current status, providing a foundation for designing more effective and adaptive group recommender systems. The findings of this research contribute to the understanding of group decision-making and offer practical implications for developing more effective and user-centered GRSs.