Trust based latency aware influence maximization in social networks Rezvan Mohamadi-Baghmolaei, Niloofar Mozafari, Ali Hamzeh n Department of Computer Science and Engineering and InformationTechnology, School of Electrical Computer Engineering, Shiraz University, Shiraz Avenue, Iran article info Article history: Received 22 June 2014 Received in revised form 21 October 2014 Accepted 9 February 2015 Keywords: Influence maximization Information diffusion Independent cascade model Trust path Distrust path TLIM abstract Influence maximization is the problem of finding a small set of nodes that maximizes the aggregated influence in social networks. The problem of influence maximization in social networks has been explored in many previous researches. They have mainly relied on similar temporal chances for every node to influence another; whereas in reality, time plays a major role in pairwise propagation rates in social networks. However, there is little research done on influence maximization considering temporal dynamics of the networks and existing approaches merely offers a mediocre performance due to ignoring trust aspects of the diffusion process. In this paper, we propose a Trust based Latency aware Influence Maximization model, abbreviated as TLIM, which selects the most influential nodes in social networks with considering time and trust simultaneously. To the best of our knowledge, we are the first to study trust in classic IC model and also the first to consider both important time and trust factors jointly in influence maximization problem. The main contributions of this paper are listed as follows: first, we extend the classic IC model to include time and trust simultaneously, which is more applicable in existing social networks. Second, we find the most influential nodes in social networks with considering time and trust together; and the last but not the least, it is applicable to well-known real social networks such as Epinions, Slashdot and Wikipedia. To explore the advantages of our approach, quite a lot of experiments with different specifications are conducted. The obtained results are very promising. & 2015 Elsevier Ltd. All rights reserved. 1. Introduction Nowadays, the rapid growth and popularity of online social netw- orking sites have brought a great deal of attention to social networks (Qiao et al., 2012; Chen et al., 2011; Barbieri et al., 2012; Lu and Lakshmanan, 2012; Belák et al., 2012; Rodriguez and Schölkopf, 2012; Ahmed and Ezeife, 2013). Beside a means of communication, online social networks provide immense sources of information, experience and innovations that enable everyone from everywhere to create, exploit, or spread content through the network via internet links. More importantly, social interaction plays a central role in shaping political or social movements and debates, such as critical role of Facebook or Twitter in the 2010 Arab Spring (Howard and Duffy, 2011). Social network analysis can help extracting worthwhile knowl- edge, controlling methods of exchanging data, maximizing acquisi- tion of information in the network and also designing improved social networks with appropriate facilities of dissemination. A lot of studies have been done in the context of information dif- fusion in social networks. Precisely, information diffusion is a research domain that concerns with the processes of dissemination of information and opinion sharing among members of a social net- work. According to a recent survey (Guille et al., 2013), studies conducted in this field include three general branches as following: “Detecting Interesting Topics”, “Modeling Diffusion Processes” and “Identifying Influential Spreaders”. The latter branch is what we study in this article, which is known as “influence maximization”. When news and innovations arise in a social network, they usually begin to spread through the network from person to person, in a virus manner, to achieve as large individuals as possible. The extent of diffusion in the network mainly relates to the mutual relationships of its members. Consider the phone network of a group of individuals. If we send a message including a rumor to someone in this network, he will inform those in his contacts about the new message. They will either accept or ignore the rumor. In fact, if they are affected by the sender, they will believe the new announcement and begin to spread it through the rest of the network using their influence on their friends. Depending on the initial receiver of the message, final amount of receivers in the network would be different. This process is called “word-of-mouth” effect in social networks and is so much operational in commercial intentions. Recently, online social communities have become the target of many companies as a way of advertising new products (Schivinski and Dabrowski, 2013). These companies aim to find the most influential individuals who are most suitable for pro- moting their brands and absorbing the most customers. They give free Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/engappai Engineering Applications of Artificial Intelligence http://dx.doi.org/10.1016/j.engappai.2015.02.007 0952-1976/& 2015 Elsevier Ltd. All rights reserved. n Corresponding author. Tel.: þ98 7136133175. E-mail address: ali@cse.shirazu.ac.ir (A. Hamzeh). Engineering Applications of Artificial Intelligence 41 (2015) 195–206