Speakers and abstracts:

Mapping flow pathways in complex systems

Martin Rosvall (Department of Physics, Umea University)

Random walks on networks is the standard tool for modelling spreading processes in social and biological systems. This first-order Markov approach is used in conventional community detection, although itRésultat de recherche d'images pour "martin
              rosvall" ignores a potentially important feature of the dynamics: where flow moves to may depend on where it comes from.

 In my talk, I will show that ignoring the effects of second-order Markov dynamics has important consequences for community detection, but also that our generalization to second-order Markov dynamics is just a first a step in the right direction, because real flow pathways contain plentiful higher-order regularities that must be disregarded. Overall, I will argue that we must move beyond the conventional focus on network topology to higher-order and multilevel network models of flows in social and biological systems, because in the wasted data are the answers to many important research questions across the sciences. And we must develop computationally efficient methods that based on the richer data can uncover the systems’ large-scale organization and changes.

Linkurious: graph visualization for all companies

Sébastien Heymann (Linkurious)

Linkurious helps organizations use their data to uncover relationships among people, places, things, and entities. It provides new capabilities to extract information from large and complex datasets.

Until recently, graph visualization was an area of interest reserved to scientists and intelligence agencies. However, as more and more organizations work with big data, they struggle to extract concrete insights from large and complex datasets. What is the impact of a server failure in a large IT network? Are seeminphotogly normal bank customers indirectly connected to known criminals? Who is the most influential person in my organization? Graph visualization can help IT engineer, fraud analysts or decisions-makers find the answers to these questions by exploring the connections in their data. However, to date, graph visualization was reserved to organizations capable of building their own solutions or using complex tools designed for scientists.

Linkurious is already helping early customers like Ebay, Cisco, or the French Ministry of Finances to extract insights from their graph data. Recently the International Consortium for Investigative Journalism (ICIJ) used Linkurious Enterprise for the Swiss Leaks, a worldwide data-investigation of a giant tax evasion scheme allegedly operated with the knowledge of of the British multinational bank HSBC via its Swiss subsidiary.

In this talk you will discover how graph visualization is used by companies of various sectors to tackle graph challenges.

The spatial structure of mobility networks (cancelled)

by Marc Barthelemy (CEA Institut de Physique Théorique)

Mobility networks described by an origin-destination matrix representing commuting flows are good examples of networks with various layers of complexity: they are directed, temporal, and weighted network, and encode much of the urban structure. The extraction of a clear and siphoto_full.jpgmple footprint of the structure of these large networks is thus an important problem for mobility and urban studies but is also very general and has many applications. I will discuss here a versatile method which extracts a coarse-grained signature of weighted, directed networks, under the form of a 2x2 matrix that separates flows into four categories. This method allows to determine categories of weighted directed networks and, applied to origin-destination matrices obtained from mobile phone data, to classify cities according to their commuting structure.

 Predicting scientific success based on coauthorship networks

by Frank Schweitzer (Chair of Systems Design, ETH Zurich)

We address the question to what extent the success of scientific articles is due to social influence. Analyzing a data set of over 100,000 publications from the field of Computer Science, we study how centrality in the coauthorship network differs between authors who have highly cited papers and those who do not. We further show that a Machine Learning classifier, based only on coauthorship network centrality metrics measured at the time of publication, is able to predict with high precision whether an article will be highly cited five years after publication. By this we provide quantitative insight into the social dimension of scientific publishing – challenging the perception of citations as an objective, socially unbiased measure of scientific success.

Reference: EPJ Data Sci. (2014) 3: 9

Revealing latent factors of temporal networks for mesoscale intervention in epidemic spread

by Ciro Cattuto (ISI Foundation)

The customary perspective to reason about epidemic mitigation in temporal networks hinges on the identification of nodes with specific features or ciro cattutonetwork roles. The ensuing individual-based control strategies, however, may overlook important correlations between topological and temporal patterns. Here we adopt a mesoscopic perspective based on dimensionality reduction techniques from machine learning: We use non-negative tensor factorization to build an additive representation of a temporal network in terms of mesostructures, such as cohesive clusters and temporally-localized mixing patterns. This representation allows to determine the impact of individual mesostructures on epidemic spread and to assess the effect of targeted interventions that remove chosen structures. We illustrate this approach using high-resolution social network data from sociometers and show that our method affords the design of effective mesoscale interventions.

Compensating for population sampling in simulations of epidemic spread on temporal contact networks

by Alain Barrat (ISI Foundation and Centre de Physique Théorique)

Data describing human interactions often suffer from incomplete sampling of the underlying population. As a consequence, the study of conAlain Barrattagion processes using data-driven models can lead to a severe underestimation of the epidemic risk. Here we present a systematic method to correct this bias and obtain an accurate estimation of the risk in the context of  epidemic models informed by high-resolution time-resolved contact data. We consider several such data sets collected in various contexts and perform controlled resampling experiments. We show that the statistical information contained in the resampled data allows us to build surrogate versions of the unknown contacts and that simulations of epidemic processes using these surrogate data sets yield good estimates of the outcome of simulations performed using the complete data set. We discuss limitations and potential improvements of our method.

Understanding Information Exchange in Social Media Systems

by Krishna Gummadi (Networked Systems Research Group, Max Planck Institute for Software Systems)

The functioning of our modern knowledge-based societies depends crucially on how individuals, organizations, and governments exchange information. Today, much of this information exchange is happening over the Internet. Recently, social media systems like Twitter and Facebook have become tremendously popular, bringing with them profound changes in the way information is being exchanged online. In this talk, I will focus on understanding the processes by which social media users generate, disseminate, and consume Krishna Gummadi's pictureinformation. Specifically, I will investigate the trade-offs between relying on the information generated by (i.e., wisdom of) crowds versus experts and the effects of information overload on how users consume and disseminate information. I will also highlight limitations of our current understanding and argue that an improved understanding of information exchange processes is the necessary first step towards designing better information retrieval (search or recommender) systems for social media.

Non-Markovian Models of Networked Systems

by Renaud Lambiotte (University of Namur)

When modelling flows in networked systems, it is often assumed that the dynamical process is Markovian. The main purpose of this work isRésultat de recherche d'images pour "renaud
              lambiotte" to test this hypothesis in a variety of systems, and to develop a theoretical framework and algorithms dedicated to systems non-Markovian in time (when I move depends on when I arrive), or in path (where I go to depends on where I am coming from).