Evaluation And Optimization Mechanisms Of Node Influence In Social Networks

Abstract:The rapid development of Internet and web technology has given rise to the rapid popularity of online social networks, such as Facebook, Twitter, Weibo. Large-scale online social networks have already started to influence people’s way of life in many ways, especially in terms of the spread of information and dissemination, attracted many scholars to study. In this paper, social networks have been studied two aspects:evaluation of the influence of nodes and influence maximization.Microblogging network exists to represent the latest online social network of high-impact nodes in the network for the dissemination of information plays a vital role. Quickly and efficiently identify these nodes help to control public opinion research, research networks of individual relationships, making the network more efficient dissemination of information. In the evaluation of the influence of the node in question, the degree centrality of the traditional method is simple, but the effect is bad; Betweenness centrality is accurate evaluation results, but the need for referrals numerical nodes based on the global information network, a large computational overhead, is not applicable in large scale networks. In this paper, based on the evaluation of the influence of ideological voting PageRank algorithm, proposed the influence of accurate and efficient sorting algorithms evaluate the influence of the nodes in the network nodes. Depending on the complexity of the relationship in online social network, we propose new propagation model different sides have different probabilities of infection spread in the SIR on the base model. Sina Weibo real user relational data set, the new sorting algorithm proposed in this paper compared betweenness centrality and PageRank algorithm has better performance.In marketing, the use of "mouth to mouth" among users of "viral marketing" program to get the attention of many scholars. How to effectively select k nodes affect the composition of the seed so that the widest range of collections is the current hot issue. Kempe proved that this problem is NP-hard, and made a greedy algorithm, with better results. Tian Jiatang proposed hybrid algorithm effectively solves the problem of time overhead with greedy algorithm. Hybrid algorithm is divided into inspire stage and greedy stage. Accumulation use linear threshold model influence, we seclect the most "potential" node in inspire stage. This paper presents optimization algorithms take advantage of the inspiration phase node neighborhood information, consider the strength of the connection node and the ability to communicate edge neighbor differences, made more effective "potential" node selection strategy. In greedy phase, optimization algorithms through each round advance to spread the seeds of the current collection, get the current active set, to avoid the collection node computing marginal revenue, greatly reducing the amount of repetitive calculations, improving the efficiency of greedy stage.