Analysis of Complex Networks

16.8 Summary

We have presented an investigation of networks derived from the European Union's Framework Programs for Research and Technological Development. The networks are of substantial size, complexity, and economic importance. We have attempted to provide a coherent picture of the complete process, beginning with data preparation and network definition, then continuing with analysis of the network structure and modeling of network formation.

We first considered the challenges involved in dealing with a large amount of imperfect data, detailing the tradeoffs made to clean the raw data into a usable form under finite resource constraints. The processed data were then used to define bipartite networks with vertices consisting of all the projects and organizations involved in each FP. To provide alternative views of the data, we defined projection networks for each part (organizations or projects) of the bipartite networks. Additionally, we used results of a survey of FP5 participants to define a smaller network about which we have more detailed information than we have for the networks as a whole.

Next we examined the structural properties of the bipartite and projection networks. We found that the vertex degrees in the FP networks have a highly skewed, heavy tailed distribution. The networks further show characteristic features of small-world networks, having both high clustering coefficients and short average path lengths. We followed this with an analysis of the community ...