NoT 2013 - The 2013 International Workshop on Networks over Time (NoT2013)

While much of network sciences research has focused on studying static networks, almost all real networks are dynamic in nature. Those networks exhibit structural changes over time. The importance of studying and modelling such networks is evident from its increasing presence : telecom call data graphs, communication networks, social (friends') networks, biological networks, language, etc. Recently, research areas for studying the topology, evolution and applications of complex evolving networks have gained more attention. In general, network modeling has long drawn on the tradition of social network analysis and graph theory. In the last decade, there is a growth in class of dynamic network models, from Erdos-Renyi model to Logit models, p*-models, and Markov random graphs. These network models, exemplified by the preferential attachment model, are widely popular in statistical physics and computer science research communities. Most of the these works deal with identifying properties in a single snapshot of a large network, or in a very small number of snapshots. The network properties include heavy tails for in-degree and out-degree distributions, communities, small-world phenomena, etc. However, in cases of availability of information about network evolution over long periods and/or at smaller granularity of time periods, for example telecom call data graphs, these methods are not very effective. Leskovec et al. in their KDD 2005 paper, point out that though models like preferential attachment are good at generating networks that match static "snapshots" of real-world networks, they do not appropriately model how real-world networks change over time. Since these models have network's topology at the core, they are well-suited for capturing the essential features of static networks (or time aggregated snapshots of dynamic networks) only. However, in many cases the interactions among the nodes are not only dynamic but also defined on a very short time scale. This has led to the development of activity driven network models. Even with these advancements, dynamic network analysis research area lacks a wholistic view of methods and models proposed in different domains, w.r.t properties and processes/activities, analyzed and modeled respectively. It is also critical to review each of the techniques for various problems and application domains, which can help in coming up with unified frameworks that can potentially address the shortcomings.