Diffusion of interests in social networks

This thesis provides a model for diffusion of interests in Social Networks (SNs). It demonstrates that the topology of the SN plays a crucial role in the dynamics of the individual interests. Understanding cultural phenomena on SNs and exploiting the implicit knowledge about their members is attracting the interest of dierent research communities both from the academic and the business side. The community of complexity science is devoting signicant eorts to dene laws, models, and theories, which, based on acquired knowledge, are able to predict future observations (e.g. success of a product). In the mean time, the semantic web community aims at engineering a new generation of advanced services by dening constructs, models and methods, adding a semantic layer to SNs. In this context, a leapfrog is expected to come from a hybrid approach merging the disciplines above. Along this line, this work focuses on the propagation of individual interests in social networks. The proposed framework consists of the following main components: a method to gather information about the members of the social networks; methods to perform some semantic analysis of the Domain of Interest; a procedure to infer members' interests; and an interests evolution theory to predict how the interests propagate in the network. As a result, one achieves an analytic tool to measure individual features, such as members' susceptibilities and authorities. Although the approach applies to any type of social network, here it was tested against two research communities in the domain of computer science and physics. The DBLP (Digital Bibliography and Library Project) database and the APS (American Physical Society) dataset were elected as test-cases since they provide the most comprehensive list of scientic productions in their respective elds.