International Journal of
Information and Education Technology

Abstract—It’s a challenging task to find a subset of node of size k in a social network such that targeting them initially as the seeds will maximize the influence spread. This problem is proved to be a NP-hard problem. We solve this problem in two aspects: 1) we improve the basic greedy algorithm, limiting the influence spread in a neighbor space to reduce the running time. We use the DAG and the recursion method to calculate the influence spread of each node. Also we transform this problem to a reachable probability query problem in an uncertain graph; 2) we present a more accurate degree discount heuristic algorithm which considers the relationship between the node and its neighbors. Intensive experiments on a large real-world social network show that: our improved greedy algorithm and degree discount heuristic algorithm are more efficient than the basic greedy algorithm and other heuristic methods.

Index Terms—Classify-tree, DAG, degree heuristic, greedy, influence spread maximization, sampling.

The authors are with the Massive Data Computing Research Lab, Computer Science and Technology, Harbin Institute of Technology (e-mail: xinfeishi@gmail.com, wangzh@hit.edu.cn, lijzh@hit.edu.cn).

[PDF]

Cite:Xinfei Shi, Hongzhi Wang, Jianzhong Li, and Hong Gao, "Influence Spread Maximization in Social Network," International Journal of Information and Education Technology vol. 3, no. 6, pp. 660-666, 2013.