A Novel Community-Based Approach for Influence Maximization in Social Network

Abstract

Our study concerns an important current problem, that of influence maximization in social network. This problem has received significant attention from the researchers recently, driven by many potential applications such as viral marketing and sales promotion. It is a fundamental issue to find a subset of key individuals in a social network such that targeting them initially (e.g. to adopt a new product) will maximize the spread of the influence (further adoptions of the new product). However, the problem of finding the most key nodes is unfortunately NP-hard. It has been shown that a greedy algorithm with provable approximation guarantees can give good approximation. However, it is computationally expensive, if not prohibitive, to run the greedy algorithm on a large social network. Based on the community structure property of social network, a cooperative game theoretic algorithm CGINA to find key nodes is proposed. The proposed algorithm encompasses two stages. Firstly, we detect the community structure of the social network with the topological structure and information diffusion model. In this paper, we adopt the algorithm 2 in [4]. Then, we will find key nodes in communities. In this paper, we think of the information diffusion in the whole network as a cooperative game with transferable utility. The communities of the network happen to be the players in this cooperative game. With the solution concept Shapley value in game theory, we allocate the number of key nodes to discover in the community. Moreover, different from the existing relevant literature, in my opinion, the key nodes two parts. One is composed of “bridge” nodes, which contact other communities densely, and are easy to disseminate information across communities, the other is composed of “influential” nodes, which are at the core of the community, in close touch with other nodes, and can diffuse information quickly within the community. To adjust the ratio of these two kinds of nodes in the same community, a heuristic factor l is introduced. Empirical studies on a large social network show that our algorithm is efficient and powerful.