科研之友
微信
新浪微博
Facebook
分享链接
摘要
The boom of social networking services makes it convenient to organize or participate in social events. Recent studies consider the social event organization problem, but they cannot provide diverse choices for users when there are multiple events. In this paper, we seek to devise an event organization scheme that provides diverse choices for users. For this goal, we explicitly distinguish the subjective preferences of users to events and the objective preferences of users to events. The former is considered to be the generative probabilities of users appearing in events, and the latter is viewed as the posterior probabilities of users participating in events. The Expectation-Maximization algorithm is employed to connect them together. After extracting these features, we extract an edge-weighted bipartite graph as the scheme from all the objective preferences. It captures the maximum sum of the objective preferences, and simultaneously satisfies soft user capacity constraints and hard event capacity constraints. We prove this problem is in P via transforming it into linear program (LP). In consideration that LP may have no feasible solutions, we alternatively devise an efficient polynomial time algorithm which can yield an approximately feasible solution. Experimental evaluation shows the effectiveness and efficiency of our proposed method.
