Abstract
Social network services to date facilitate new ways of communication and generate huge amounts of user data, providing better personalization and direct marketing for individual users. However, with respect to group functionality social network services are still stuck in the days of online forums and Usenet groups. Here we propose a new approach, where a user group can be organized dynamically into subgroups to plan common real-life activities. Bestmatching subgroups are determined on the basis of individual user preferences as well as subgroup formation preferences from a group administrator. The latter can adapt multi-disciplinary findings on group formation. Moreover, our approach can cope with the phenomenon of data adaptivity. We have built a first prototype by using Preference SQL. Due to its high combinatorial complexity, solutions inspecting the complete search space are only feasible for small user groups. Preliminary benchmark results show that indeed for small groups best-matching subgroup formations can be determined in a reasonable amount of time. Heuristic preference query optimization techniques for larger group sizes are a big challenge for future work.
Authors
Florian Wenzel, Werner Kießling
Publication
Proceedings of the 6th International Workshop on Personalized Access, Profile Management, and Context Awareness in Databases (PersDB ’12) in conjunction with VLDB, 27-31 August, 2012, Istanbul, Turkey