Effective virtual communities appear to be an exceptional approach, which bring together people from diverse background and provide support for knowledge sharing. A common misconception though, is to believe that a VC will be effective when people and technology are present. This work examines computational means for providing personalised support tailored to the needs of a VC rather than supporting individual members Based on 4 processes identified as important for successful knowledge sharing in closely-knit communities, we propose a computational framework that includes the extraction of a community model and the deployment of that model to provide support adapted to the whole community. This paper will present the initial formalization of the input data, along with an initial attempt to define relationships and centrality of community members.