In this project we want to develop statistical models to describe networks growth and evolution. These models will be based on agents interactions and inspired to the theory of Self-Organization and Fractal Growth. At the same time we are thinking to collect data mainly for Internet and World Wide Web structure in order to validate the models and we want to devise visualization tools in order to analyze large data sets both from numerical simualtions and from real-world data. Application to social andconomic networks will be also considered.
DESCRIPTION OF WORK
(The study of the dynamics of complex networks requires an interdisciplinary approach that will use methodologies developed in different areas. Complex Systems theory is the natural candidate framework for the study of such networks. It provides a unified language and a set of operative tools to address the fundamental issues involved in the study of such systems. In this framework we want to devise common methodologies for running empirical simulations, computing the relevant quantities on the simulated models and on real samples and visualizing the evolution of networks even on very large dimensions.
The unifying feature of these networks is that their global structure and dynamical evolution are the result of locally interacting agents distributed in the system. Shape of the network and agent requests co-evolve to form the final structure. We plan to devise new stochastic models for Internet to be validated through an extensive comparison with a multiview observation from different locations. We also want to devise content sensitive stochastic models of the world Wide Web where documents link to other documents that are relevant for a common subject of interest. The study of such large networks require the development of ad-hoc visualization algorithms and data structures for drawing and browsing virtually infinite structures rapidingly evolving in time.
We shall also consider the analysis of social and economic networks as cyber communities, correlations between stocks and firm etc. We finally want to test the existing models for the Internet and the World Wide Web by comparing the behaviour of suitable quantities identified through theoretical analysis and large computer simulations. In particular we refer to the various frequency distributions observed in real samples as the degree distribution, the number of hops required to reach boundary of the system etc.
Funding SchemeCSC - Cost-sharing contracts
75230 Paris Cedex 05