Fast Budgeted Influence Maximization over Multi-Action Event Logs

Abstract

In a social network, influence maximization is the problem of identifying a set of users that own the maximum {\it influence ability} across the network. In this paper, a novel credit distribution (CD) based model, termed as the multi-action CD (mCD) model, is introduced to quantify the influence ability of each user, which works with practical datasets where one type of action could be recorded for multiple times. Based on this model, influence maximization is formulated as a submodular maximization problem under a general knapsack constraint, which is NP-hard. An efficient streaming algorithm with one-round scan over the user set is developed to find a suboptimal solution. Specifically, we first solve a special case of knapsack constraints, i.e., a cardinality constraint, and show that the developed streaming algorithm can achieve ($\frac{1}{2}-ε$)-approximation of the optimality. Furthermore, for the general knapsack case, we show that the modified streaming algorithm can achieve ($\frac{1}{3}-ε$)-approximation of the optimality. Finally, experiments are conducted over real Twitter dataset and demonstrate that the mCD model enjoys high accuracy compared to the conventional CD model in estimating the total number of people who get influenced in a social network. Moreover, through the comparison to the conventional CD, non-CD models, and the mCD model with the greedy algorithm on the performance of the influence maximization problem, we show the effectiveness and efficiency of the proposed mCD model with the streaming algorithm.