On this page, we list the major methodological challenges that will be addressed by HumanICT and explored through specific applications described in the ”Domains & Applications” section.
Today's computers can easily perform simulations of large numbers of interacting agents. Well motivated models of agents in interactions may generate unexpected collective effects which call for mathematical understanding. Conversely, collective effects predicted from models simple enough to be mathematically analyzed call for “empirical” (numerical) tests on more complex (more realistic) multi-agents systems.
With this general goal of understanding and anticipating behaviors of complex social systems, one needs to develop simplified models, on which the mathematical analysis can be done. Such models should be able to reproduce the stylized facts empirically observed. From the analysis of these models, new intuition can be gained, and more complex models can be studied. Scenarios can be explored and tested by the large scale simulations, providing results that can be used for decision making.
On the other side, it is also needed to develop specific mathematical and software tools to explore the dynamical behavior of computer models and to help formalising the properties of observed collective effects. Moreover, it is necessary to develop specific mathematical tools that help evaluate the uncertainties around the results. This is essential for insuring the quality of simulations.
Obviously, models should be grounded in empirical data. Calibration of models, and tests on real data, must be at the core of the modelling approach (see also next section). It will be mandatory to build or have access to large scale data bases on specific issues, but also to produce new data by making large scale surveys or devising experiments.
Due to the development of new technologies and of sensors of various types, and through the use of digital media and computational devices, we increasingly leave behind digital traces of our daily activities. The scale at which such data can be gathered and analyzed affords a novel, data-driven approach in the investigation of various aspects of human behavior. In particular, it becomes feasible to gather data on human interactions and contacts with high temporal and spatial resolution through the deployment of lightweight measuring infrastructures such as the RFID badges developed by the SocioPatterns project (www.sociopatterns.org). The patterns of human interactions can thus be investigated as dynamical networks of contacts in various environments, and with several applications.
In most applications of the project, the simulations will involve a population of agents representing the inhabitants of a city, a region, country or even the whole Earth. The level of details describing these individuals will depend on the applications. However, one can guess that basic attributes such as age, occupation and maybe education can be informative in many contexts. Moreover, it is of course crucial to get the evolution of this population in time (with different time scales and considering different scenarios). It must be stressed that one is here interested in a statistical approach: the simulated individuals will NOT correspond to any living people, to ensure that their privacy is respected, but the statistical characteristics of the virtual populations will have to match the observed statistical data at various scales.
The models give the possibility to anticipate possible futures. They can also help to define policies of action that favour viability and resilience. Designing a precise conceptual framework for these notions is important both for the discussions with the stakeholders and designing computer assisted tools for policy design.
Including stake-holders in the modelling process is generally very important. It is a challenge to use new ICT such as virtual realities and avatars to enlarge such a participation.
The project will build progressively (in collaboration with FuturICT), a platform including a distributed set of models, model results and data sources, continuously updated, that will be available for other researchers and final users. It is a challenge to define the appropriate ICT architecture, making possible the collaboration of a large number of scientific and expertise groups.
Another ICT challenge will be to develop the appropriate user interface for discussing with the platform. Indeed, this interface will have to organise the access to a large variety of data sources and simulation results, depending on possibly very complex scenarios.