In social network analysis, the detection of communities composed of people with common interests is a classical problem. Moreover, people can somehow influence any other in the community, i.e., they can spread information among them. In this paper, two models are proposed considering information diffusion strategies and the identification of communities in a scientific social network built through these two model concepts. The maximum flow-based and the Erd s number-based models are proposed as a measurement to weigh all the relationships between elements. A clustering algorithm (k-medoids) was used for the identification of communities of closely connected people in order to evaluate the proposed models in a scientific social network. Detailed analysis of the obtained scientific communities was conducted to compare the structure of formed groups and to demonstrate the feasibility of the solution. The results demonstrate the viability and effectiveness of the proposed solution, showing that information reaches elements that are not directly related to the element that produces it.
展开▼