The Clique research group in University College Dublin is focused on the analysis and visualisation of social networks. Computer scientists and computational statisticians are working together on problems including community-finding in social networks, influence propagation and detection of anomalous structure in networks. Research is driven by the analysis of large-scale networks provided by industrial partners, in particular, networks of mobile phone-calls containing more than a million nodes and tens of millions of links. In this talk, I will focus primarily on the community-finding problem, discussing initially the structure of real-world networks and on how this impacts on the communities that likely to be found in such networks. I will argue that the view of social networks as consisting of well-separated communities connected by weak links does not hold in many real-world networks and I will introduce algorithms that we have developed to detect overlapping community structure in networks with pervasive overlapping community structure.
Neil J. Hurley received an M.Sc. in mathematical science from University College Dublin (UCD), Dublin, Ireland, in 1988. In 1989, he joined Hitachi Dublin Laboratory, a computer science research laboratory at the University of Dublin, Trinity College,from which he received the Ph.D. degree in 1995, for his work in knowledge-based engineering and high-performance computing. He joined the academic staff of UCD in 1999 where his present research activities lie in the areas of large-scale network analysis, robust information retrieval and data-hiding.