In recent years, services such as Meetup, Plancast, and Eventbrite have provided platforms for planning and organizing live events. Social event organization (SEO), the problem of finding an assignment of users to events by considering their interests and social connections, is a problem that has received growing attention after being introduced by [24]. Given a weighted bipartite graph specifying the interest of every user in every event, and a social network between the users, the main goal of the SEO problem is to assign users to events so as to maximize a social welfare function, while respecting the minimum and maximum cardinality bounds associated with events.

The problem is shown to be NP-complete, and in fact, hard to approximate [24]. First, we review the previous solutions proposed in [24] and discuss some problems in these algorithms. Then, we propose the Second-Chance Dynamic Greedy (SCDG) and Community-Aware Static Greedy (CASG) algorithms to enhance the quality of the results produced by the existing algorithms. Our experiments using both synthetic and datasets from Meetup and Plancast show that our algorithms obtain a social welfare up to 60% better than that obtained by Phantom-Aware Dynamic Greedy (PADG), the best algorithm in [24].

Second, we propose the personality-oriented objective function that takes into account a user’s willingness participate in large events, or events with unknown people. We adapt PADG, SCDG, and CASG so as to optimize this new objective function. Our experiments show that the personality-oriented version of SCDG improves the social welfare by up to 100% over the adaptation of PADG.