GENeSIS: Generative E-Social Science

The GENESIS Project is a unique collaboration between two groups of social scientists at the University of Leeds and University College London whose expertise is in simulating the spatial dynamics of development in cities and regions using large scale data and computational resources. Fifty years ago, typical social science models were highly aggregate.

They treated the social system as if it were in equilibrium and articulated most social processes in an orderly manner from the top down. In the last twenty years, models of social systems have undergone a sea change not only in response to much finer scale disaggregate data and dramatic increases in computational resources but largely because the predominant way in which society is now conceived is a mixture of bottom-up with top-down processes.

As a result, new styles of modelling making use of these radically enhanced computational and data resources and representing social systems as groups of individuals and institutions have come onto the agenda. Such models can generate surprising patterns which 'emerge' from the operation and interaction of individual behaviours. Such simulations are often referred to generically as 'agent-based models'.

They are capable of addressing many more questions than previous generations of model and have important applications in policy analysis where unanticipated outcomes are the norm. As these models focus on processes rather than outcomes per se, they are essentially generative in their structure, building on developments in evolutionary theory, complex systems, and new approaches to dynamics.

They are increasingly dependent on new computational infrastructures such as those being pioneered in the National Centre for e-Social Science. The GENESIS Project (GENerative E-SocIal Science) is centrally focussed on exploiting this infrastructure for the development of this new generation of simulation models and policy applications.

The two groups, the existing MoSeS and GeoVUE nodes in NCeSS, will develop a series of generic and specific demonstrator simulations based on a synthesis of agent-based modelling, microsimulation, and visual simulation. These models are focussed on residential and retail-services location which are the core components of the demographic and economic sectors of the urban and regional system.

They form the core of a comprehensive structure for simulation into which more specific demonstrators dealing with local movement (congestion and crowding), residential segregation, and demographic aging and lifecycle effects (particularly in terms of income and welfare) can be plugged. 

These demonstrators provide a basis on which other users can elaborate and tune the simulations to their own applications. The project will establish this computer environment through the current e-infrastructure portal and related services which are being built at NCeSS, with customised portlets being designed to accommodate specific applications.

Access and outputs from the system will be strongly visual with substantial visual iconography being used to enable users to apply and tune their own applications within the environment, and outcomes from the models being available using non-proprietary webased services such as 2D and 3D mapping which contain effective multimedia for scientific visualisation.

The existing set of methods and techniques being developed by MoSeS for microsimulation using reconstructed population profiles with dynamic aging and by GeoVUE for visualisation using the MapTube software products will form the basis for the extended simulation platform developed here. 

The project is supported by key partners: Google, ESRI, GLA Economics, and the British Geological Survey, and it has strong international collaboration with groups at Los Alamos/Santa Fe Institute, George Mason and Michigan Universities, the University of Paris 1 as well as other partners in the NCeSS consortium.

People

• Andrew Hudson-Smith
• Anders Johansson
• Richard Milton
• Oliver O'Brien
• Kiril Stanilov

Outputs

Batty, M. Carvalho, R., Hudson-Smith, A., Milton, R., Duncan Smith, D., and Steadman, P. (2008) Scaling and Allometry in the Building Geometries of Greater London, European Physical Journal B, 63, 303-318.

Batty, M., Hudson-Smith, A., Milton, R. and Crooks, A. (2010) Map Mashups, Web 2.0 and the GIS Revolution, Annals of GIS, 16(1), 1-13.

Hudson-Smith, A., Crooks, A., Gibin, M., Milton, R., and Batty, M. (2009) NeoGeography and Web 2.0: Concepts, Tools and Applications, Journal of Location Based Services, 3, 118-145

Hudson-Smith, A., Crooks, A., Milton, R., and Batty, M. (2009) Mapping for the Masses: Accessing Web 2.0 Through Crowdsourcing, Social Science Computer Review, 27, 4, 524-538.

Lin, H. and Batty, M. (Editors) (2009) Virtual Geographic Environments, Science Press, Beijing, China, vi + 350 pp. (with H. Lin)

Stanilov, K. and Batty, M. (2011) Exploring the Historical Determinants of Urban Growth Patterns through Cellular Automata, Transactions in GIS, 15(3), 253-271

Impact

 

The GENESIS project merges the activities of two existing nodes in the National Centre for e-Social Science - MoSeS at Leeds University whose objective is to develop microsimulation for social scientists across different spatial scales, their current focus being on simulating population structure and dynamics for all households in the UK, and GeoVUE at UCL whose focus in on the development of visual environments for social science, specifically building on 2D and 3D mapping technologies.

The following general objectives define the impact of the project and these are:

• To develop for the first time in the UK, methods, data and e-infrastructure for generative social science which will include advances in agent-based models and their aggregation across spatial scales.
• To link e-social science to the emerging sciences of complexity, building on developments in agent-based modelling (ABM) and microsimulation.
• To generate new computational platforms integrating desktop, web- and grid-based environments, creating shared Collaboratories for the application of ABM by social scientists in research and policy analysis.
• To exploit, demonstrate, and impress the advances being put in place by ESRC, JISC, and other agencies involving large data sets as under the Census and National Data Sets Strategy programmes.
• To provide capacity through NCeSS for e-social science which is both usable and sustainable for the wider research community.

We realise this vision through:

• A series of generic demonstrators dealing with different sectoral simulations of the urban and regional system from local to national scales, focussing on residential and retail structure and dynamics as a precursor to more comprehensive dynamic urban simulations.
• A series of specific demonstrators focusing upon congestion and crowding at local scales, residential segregation and differentiation at neighbourhood scales, and demographic, life cycle and income effects at more aggregative scales. These will reflect key issues of contemporary social and economic policy in cities and regions.
• A set of interlinked visual environments building on non-proprietary web-based services such as those developed by Google and ESRI, two of our key partners, as well as developments in GIS, multimedia, virtual worlds, and related techniques of visualisation.
• Advances in the dynamics of geodemographics which enable rapid prototyping of new classifications through time related to the various demonstrators that we will develop.
• The NCeSS e-infrastructure platform, specifically through the general portal and specific portlets tailored to enable customisation of our demonstrators.