The Study Of Social Network Influence Maxi-mization Algorithm And Propagation Models

In recent years, the rapid development of Internet provide a large amount of data to the research of social network, while the popularity of the PC resolving the problem of processing large-scale datasets, which makes the study of social networks have made great progress. The study of social networks has many research directions such as the problem of community detection in networks, the mutual influence of nodes, nodes classification problem and so on. We focused on the influence maximization of social network. It is mainly used in business "viral marketing". The purpose is to exchange the greatest return in a limited budget. The objective of the influence maximization problem is to find an initial collection can achieve the above objective.In this paper, we propose an improved algorithm based on the greedy algorithm and an improved propagation model based on the independent cascade model. This paper studies the following questions:(1) describes the development status of the influence maximization algorithm, describes the IC and LT propagation models and influence maximization algorithms in detail, explore the advantages and disadvantages of each algorithm;(2) proposes an improved node selection algorithm based on evenly divided network, the algorithm evenly divides the network during the process of traverse network to reduce network scale, and proved by experiments that the algorithm improves the operational efficiency;(3) proposes an improved independent cascade model based on floating activation probability, the model constantly revises the activation probability in the information dissemination process to make it more in line with the actual communication process and it is proved by experiments that results of the propagation in this model is more in line with the reality results.