Research On Maximizing The Influence Diffusion In Online Social Network With Budget Constraint

With the widely usage of web social media such as Wechat and QQ, Online Social Networks becomes a new tool for information diffusion. Influence Dif fusion Maximization problem looks for a subset with k users in the network through whom the information can be diffused to the widest range.In real Influence Maximization problems, information must be diffused to people with some certain characteristics, and the budget for information diffusion is limited. To solve this problem, We use the social relations to mine user's characteristics, and design a new information diffusion model to maximizing the influence while limiting the costs within the budget.In this thesis, we design a technology to analyse the weight of different characteristics for users. In this technology we use the online social network to build a PageRank-like vote model, which is used to compute the weight of user's different characteristics. The weight of a characteristics means the importance of this characteristics, which is used to value the diffusing effect and decide a user as a target user or not.Existing information diffusion models in online social network always cost too much time to solve the influence dif fusion maximization problem. To improve the models' efficiency, this thesis proposes an improved independent cascade model which is called CGIC. By compacting the social networks using the probabilistic characteristic of IC model, the CGIC model is more time-effective.In many cases, information diffusion should be conscious with a budget constraint.We propose a budget greedy strategy called BTIDM to obtain the maximizing influence effect while limiting the costs within the budget.Based on the co-authorship relations from the arXiv website, we imply our experiments. The results show that our strategies and models are efficient.