Choice making support for a group organizer: the case of restaurant recommendation

Group recommender systems (GRSs) are designed to assist groups of users in finding products and services that best match the combined preferences of the group members. Most GRSs are designed to involve all the group members in the user/system interaction. However, in some situations, a single member may actually interact with the system and make the decision on behalf of the entire group; for instance, choosing a restaurant suited for a group to have dinner. This specific scenario has been overlooked by existing GRSs. Here, we explicitly consider a group recommendation scenario where only a single group member, the organizer, interacts with the GRS to identify and select a suitable restaurant for his/her group. In such a scenario, to make a decision suitable for the group, the organizer needs to: (i) anticipate the other members’ food preferences, to find restaurants that offer the desired dishes, and (ii) address the problems raised by incompatible group members’ preferences, which may exclude the existence of a restaurant that meets all the group members’ preferences. To support these tasks, we have designed a system prototype, which is called MyFoodGRS. It addresses some of the functional requirements arising when a group organizer is making a good choice for a group: by supporting the organizer in anticipating preferences, entering them into the system, managing incompatible preferences of group members, and finally selecting a restaurant. We have conducted a user study and several off-line experiments. Our analysis shows that the implemented decision support functionalities, enables group organizers to select restaurants that satisfy the group members.