An exploration of broader influence maximization in timeliness networks with opportunistic selection. (March 2016)

• Record Type: Journal Article
• Authors: Han, Meng; Yan, Mingyuan; Cai, Zhipeng; Li, Yingshu
• Is Part Of: Journal of network and computer applications. Volume 63(2016)
• Page Start: 39
• Page End: 49
• Publication Date: 2016-03
• Subjects: Social networks -- Influence maximization -- Algorithm -- Data mining; Microcomputers -- Periodicals; Computer networks -- Periodicals; Application software -- Periodicals; Micro-ordinateurs -- Périodiques; Réseaux d'ordinateurs -- Périodiques; Logiciels d'application -- Périodiques; Application software; Computer networks; Microcomputers; Periodicals; 004.05; 004
• DOI: 10.1016/j.jnca.2016.01.004
• Languages: English
• ISSNs: 1084-8045

Abstract

Abstract: The goal of classic influence maximization in Online Social Networks (OSNs) is to maximize the spread of influence with a fixed budget constraint, e.g. the size of seed nodes is pre-determined. However, most existing works on influence maximization overlooked the information timeliness. That is, these works assume that the influence will not decay with time and the influence could be accepted immediately, which are not practical. Second, even the influence could be passed to a specific node in time, whether the influence could be delivered (influence take effect) or not is still an unknown question. Furthermore, if let the number of users who are influenced as the depth of influence and the area covered by influenced users as the breadth, most of research results only focus on the influence depth instead of the influence breadth. Timeliness, acceptance ratio and breadth are three important factors neglected before but strongly affect the real result of the influence maximization. In order to fill the gap, a novel algorithm that incorporates time delay for timeliness, opportunistic selection for acceptance ratio, and broad diffusion for influence breadth has been investigated in this paper. In our model, the breadth of influence is measured by the number of communities, and the tradeoff between depth and breadth of the influence could be balanced by a parameter φ . Empirical studies on different large real-world social networks show that high depth influence does not necessarily imply broad information diffusion. Our model, together with its solutions, not only provides better practicality but also gives a regulatory mechanism for the influence maximization. It also outperforms most of the existing classical algorithms.