Section: Overall Objectives
The main goal of the DANTE team is to lay solid foundations to the characterization of dynamic networks, and to the field of dynamic processes occurring on large scale dynamic networks. Several kinds of real-world networks can be represented by graphs in which vertices represents entities like individuals in a social network, computers or routers for the Internet map, documents for the web, articles for citation networks, proteins or genes for biological networks, words or phonemes for linguistic networks, to mention only few examples. Interactions or relations between those entities are represented by edges. In order to develop tools of practical relevance in real-world settings, we propose to ground our methodological studies on real data sets obtained through large scale in situ experiments. Only recently it has become possible to study large scale interaction networks, such as collaboration networks, e-mail or phone call networks, sexual contacts networks, etc. This has prompted many research efforts in complex network science, mainly in two directions. First, attention has been paid to the network structure, considered as static graphs. Second, a large amount of work has focused on the study of spreading models in complex networks, which has highlighted the role of the network topology on the dynamics of the spreading. However, the dynamics of the networks, i.e., topology changes, and in the networks, e.g., spreading processes, are still generally studied separately. There is therefore an important need developing tools and methods for the joint analysis of both dynamics.
The DANTE project emphasizes the cross fertilization between these two research lines which should definitively lead to considerable advances. The DANTE project has the following fundamental goals:
To develop the study of dynamic interaction networks, through the design of specific tools combining graph theory and stochastic process targeted at characterizing and modeling their dynamic properties.
To infer from models, some statistical properties (dependencies, correlations) and some stochastic descriptions (transition law, large deviation) in order to characterize the dynamic behavior of the studied systems (non stationarity, burstiness, non-persistence).
Most activity on complex networks has up to now focused on static networks, the characterization of their structure, and the understanding of how their structure influences dynamic processes such as spreading phenomenon. The important step that the DANTE project wants to undertake is to consider that the networks themselves are dynamic entities. Their topologies evolve and adapt in time, possibly driven by or in interaction with the dynamic process unfolding on top of it.
The DANTE project therefore addresses both very fundamental and very applied aspects that are tightly linked. On one hand, to develop knowledge in the networking field, in order to provide a better understanding of dynamic graphs. This fundamental work is grounded on real world large scale dynamic networks. On the other hand, to help develop a better understanding of the physical objects and networks that are studied. This point requires the joint study of both dynamics of the network and in the network, and requires a tied collaboration with the research disciplines where the objects come from.
The impact of the research developed in DANTE goes beyond the context of spreading process / epidemic diffusion, thanks to the inherent interdisciplinary of the complex networks research field. Dynamic processes on dynamic networks are indeed present in numerous fields, including rumor spreading in social networks, opinion formation, fashion phenomena, the innovation diffusion in a population, etc. The spread of computer viruses may take place through email networks or bluetooth connections, which are both dynamical. The development of efficient algorithms for information spreading in wireless/P2P/DTN networks should also be improved by the understanding of the dynamics of these networks and their temporal properties. The study of all these processes should benefit from the tools developed in this project. It represents an important opportunity to study real-world dynamic processes occurring on interaction networks whose dynamics can be measured.