Information Diffusion Model

NLINE social networks have gained popularity among the young and the old alike in the last decade. They have played a huge role in creating the highly connected world that we live in today. But gone are the days when social networking sites were merely used in order to keep in touch with friends and family. Now, businesses use it as a source of information that can provide them with insights into how they can improve their marketing strategies. One such nugget of information is the way in which information spreads through the network. This information can be used by businesses to determine the best way to promote their products, services or campaigns. Using the model, the number of people reached can be estimated and by running it over different

Explanatory models simply try to show how the information spread through the network in the past. Predictive models, on the other hand, attempt to predict the flow of information in the future. Predictive models can either be graphical or non-graphical. Both approaches suffer from certain drawbacks and there is no way to say that one is better than the other with any degree of certainty and depends on the specific implementation.
II. LITERATURE SURVEY
A. Information Diffusion Modelling
There are two major models of information diffusion. They are linear threshold and independent cascades. A linear threshold model is a receiver centric model where each node is associated with a threshold and passes on the information only when the incoming influences exceed the threshold. An independent cascade model, on the other hand is sender centric. Each node that has obtained the information attempts to activate its neighbours once.
F. Wang, H. Wang and K. Xu [2] proposed a partial differential equation to model both the temporal and topological dynamics of diffusion of information that is injected into the network. But topology is considered only in terms of distance from the source.
A. Guille and H. Hacid [3] proposed another asynchronous information diffusion model called T-BaSIC (Time Based Asynchronous Independent Cascades). The various parameters considered are functions of time.

and Kuri, J. [4] modified the traditional SIS model to compensate for active recruitment process thus explaining the effect that advertisements have on the information diffusion process. Campaigning on a fixed budget is studied.
Jiang, Chunxiao, Yan Chen, and KJ Ray Liu [5] proposed a graphical evolutionary game theoretic framework to model and understand the information diffusion process in online social networks (both uniform degree and non-uniform degree networks). Experiments were conducted and results were verified on both synthetic and real world networks. In a related paper published by Xuanyu et al [6], the authors also study two different models of information diffusion. Based on whether the users have in depth knowledge of other users in the network, the diffusion process may be type dependent or type independent.
Garg, S. and Kumar, S [7] used the SI model to study the information flow. A number of infected nodes of a social network in NetLogo environment is taken and simulated for a different size and different average degree network taking input 100 times which helps in correction of data and improves the accuracy of the