Complexity, Networks, Geosimulations UNIL | Université de Lausanne

General Assembly European Research Group S4

Prof. Denise PUMAIN, Prof. Stephane JOOST, Dr Christian KAISER, Prof. Jean-Paul FERRIER, Prof. Jean-Bernard RACINE, Prof. Micheline COSINSCHI

Prof. David LANE, University of Modena & Reggio Emilia, Italy. For a long time, the idea of novelty was generally regarded with great suspicion; in the last few centuries, it is increasingly viewed in a favourable light, so much so that it seems appropriate to characterize the developed and developing world in the last few decades as "The Innovation Society". In this talk, I describe some aspects of social, technological, economic and cultural innovation processes, particularly those that link transformations in artefacts, attributions and social organization in a positive feedback dynamic, and the ideology that justifies and shapes the way in which this dynamic is currently playing out. I then raise some issues concerning the sustainability of this dynamic.

Prof. Lena SANDERS, CNRS, Paris, France. Cities evolve at different rhythms, some growing, other stagnating or even declining. The hypothesis is that these differences are the result of different abilities to valorise a position in a system of cities. The interactions between cities play then a driving role in their evolution. First I will briefly present a family of models based on that principle, the SimPop models. Then I will use particular one, the EuroSim model which concerns the evolution of the European cities during the period 1950-2050, in order to show and to discuss how the cities’ networks are built up. According to its economical specialisation, a city can be involved in one or several networks which are evolving through time according to a combination of exploration and exploitation principles. At last a comparison of the consequences of different scenarios (according to the existence or not of border constraints on the exchanges between cities of different countries) by 2050 is presented.

Prof. Peter ALLEN, Cranfield University, United Kingdom. Many spatial models are really ‘fitted descriptions’ of spatial changes that have occurred. In order to have a deeper and possibly more long term understanding of what is going on, or may happen, it is necessary to represent the behaviours, interactions and circumstances of the different agents within the system and to explore not only how their locations and size may change, but also how their activities, technologies, needs and requirements may change over time. This presentation looks at how complex systems models attempt to do this, and how this led to ‘multi-agent’ models several decades ago. The different types of agent are distributed spatially across the transport and communication networks of a region, both responding to, and shaping these over time. The models then show different possible patterns of spatial organization that can emerge, and allow us to see how stable different possible trajectories might be. Such models can form the basis of a ‘learning’ community, city or region where modellers, institutions, organizations and firms try to experiment and learn together about the possibilities that are open to them. This underlines the problems of finding successful levels of evaluation for decisions and policies and making operational models that reflect the spatial realities of subsidiarity for different phenomena and issues.

Prof. Stefano BATTISTON, System Design, ETH Zurich, Switzerland. We propose a novel trust metric for social networks which is suitable for application to recommender systems. It is personalised and dynamic, and allows to compute the indirect trust between two agents which are not neighbours based on the direct trust between agents that are neighbours. In analogy to some personalised versions of PageRank, this metric makes use of the concept of feedback centrality and overcomes some of the limitations of other trust metrics. In particular, it does not neglect cycles and other patterns characterising social networks, as some other algorithms do. In order to apply the metric to recommender systems, we propose a way to make trust dynamic over time. We show by means of analytical approximations and computer simulations that the metric has the desired properties. Finally, we carry out an empirical validation on a dataset crawled from an Internet community and compare the performance of a recommender system using our metric to one using collaborative filtering.