Network modeling
Artificial networked systems can significantly improve infrastructures. The initial impression of randomness in the formation of these networks has been replaced by the systematic look and study for regularities and patterns which can be expressed statistically. The development of statistical models to describe networks growth and evolution were the subject of COSIN project - Co-evolution and self-organization in dynamical networks. These models are inspired by the theory of Self-Organisation and Fractal Growth and are based on structures originated by the interplay of different agents in information society. As systems may be represented by graphs describing various interactions among their parts, WWW can be considered as WebGraphs where nodes are static html pages and edges are hyperlinks between these pages. These graphs have been the subject of extensive attention in an effort to understand the structure of WWW and consequently bring benefits to large scale web applications. In order to model network evolution, large datasets and massive graphs must be evaluated, which yield a mapping of the dynamical growth process. To this end, the COSIN network of partners developed a collection of algorithms and related implementations to generate and analyse Web-graphs. The resulting library contains routines able to perform measurements on large graphs, which can be computed on a medium size PC. The analysis of the link structure of the Web is important for applications ranking Web documents returned by a query to a search engine, such a Page Rank. The library contains routines for simulating models of stochastic graphs resembling the properties of the Web. This is carried out by measuring the Page Rank, finding correlation between different measures, and connected components. Additionally, the routines may be also used for tracing the emergence of hidden cyber-communities and detecting the overall picture of the structure of the Web-graph. The library is publicly available and it is used by different research groups in Europe as it comprises a complete suite of routines for analysing large Web-graphs.
