Study of Diffusion Models in an Academic Social Network

Models for the processes by which ideas and influences propagate through a social network have been studied in number of domains, including the diffusion of medical and technological innovations, the sudden and widespread adoption of various strategies in game-theoretic settings, and the effects of ’

word of mouth

’ in the promotion of new products. The problem of selecting a set of most influential nodes in a network has been proved to be NP-hard. We propose a framework to analyze the network in depth and to find the set of most influential nodes. We consider the problem of selecting, for any given positive integer

k

, the most influential

k

nodes in a Academic Social Network (ASN), based on certain criterions relevant in academic environment like number of citations, working location of authors, cross reference and cross co-authorship. Based on the initial node set selection and the diffusion model, we study the spread of influence of the influential nodes in the academic network. Appropriate criterions are used in the proposed generalized diffusion models. In this paper, we used two different models; (1) Linear Threshold Model and (2) Independent Cascade model to find the set of influential nodes for different criterions in an ASN and compared their performances. We constructed ASN based on the information collected from DBLP, Citeseer and used Java Social Network Simulator (JSNS) for experimental simulations.