Research lab (Forschungspraktikum)

The research lab aims at integrating methods of data driven analysis with conceptual modeling for social simulation. Data collection and analysis are directed towards data originating from social media (like Twitter, Facebook, Instagram…), taking advantage of the embedded meta-data for social network analysis. The key research question is to evaluate how far automated data analysis can support (semi-)automatic conceptualization of a simulation model (given the need for traceability of provenance information in the simulation model).

Informationen aus KLIPS

    Organisational issues

    • Kick-off meeting: the date will be coordinated as soon as 6 participants are registered in Klips.
    • The kick-off, regular (approx. bi-weekly) and internal (student team)
    • Target audience: Master E-Government, Master Information Systems, Master Information Management, Master Web Science, Master Computer Science
    • Description of the module, including workload: see WebSis: 04FB2003
    • Language: mixed English and German (depending on Students)
    • Examination: Regular and active participation in project meetings, including presentations and performance of specific project tasks, final presentation, documentation of results in the project handbook
    • Registration in Klips is open.

    Background and scope

    Combining computational methods and social sciences is a hot topic both in academic research and for practitioners. On the one hand, the overwhelming availability of sources of quantitative data puts data-driven analysis at a vantage point, requiring adequate means and methodologies for an appropriate analysis. On the other hand, data can be used in simulations to attempt understanding processes in retrospect or regarding potential alternative scenarios of the future, gaining insight to micromotives resulting in emerging macrobehaviors, all hidden in the data themselves.

    Hence, it is important to find processes and tools to tightly integrate data analysis methods and simulation, to ignite a virtuous cycle of observation, conceptualization, implementation, and back. Previous research – see e.g. the OCOPOMO (http://ocopomo.eu) and GLODERS (http://gloders.eu) projects –has shown how stakeholder participation and qualitative text analysis can be coupled with agent-based simulation models [1]. Applying this process, information elicited from unstructured texts together with insights from discussions with stakeholders could be transformed into consistent conceptual descriptions (CCDs, using the CCD tool developed at the University of Koblenz [2]). From such CCDs, agent-based simulation models were developed applying a model-driven transformation approach [2].

    This combination of data, analysis techniques, CCD, simulation model and –  ultimately – simulation results is grounded on a traceability approach, a conceptual and technical means to trace simulation results back to the provenance constituted in the evidence data [3][4].  

    The aim of the research lab is to extend the conceptual modeling process with data collection and analysis techniques for data originating from social media (like Twitter, Facebook, Instagram…), taking advantage of the embedded meta-data for social network analysis. The key research question is to evaluate how far an automated analysis and conceptualization of a simulation model is possible with this kind of data, and how the traceability feature can be supported.

    The research lab builds upon the existing body of knowledge, tools developed and mix of competences available at the research group E-Government.

    Students shall work along the work packages described below. Seed data on top of which the research lab will revolve may be provided either in-house (other projects lend themselves to the purpose), by students, or other third parties (depending on the policy case to investigate).

    Suggested Work Packages

    1. Define a use case: (a policy question where data from social media matters, such as phase-out of usage of coal as energy source, separation of the University locations of Koblenz and Landau, or other proposals by students that can be discussed at the kick-off meeting).
    2. Identify the relevant stakeholders (individuals: citizens, unionists, students...; organized groups: grassroots movements; media outlets; politicians: individual lead roles, political parties) and the relevant communication channels they use – both among themselves and beyond group borders. Describe the kind of information that can be gathered from these stakeholders and collect the data, e.g. Twitter streams or “scenarios” from politicians. In the case of Twitter data a topic of interest can be readily expressed as a set of hashtags and/or keywords: “GMO”, “hybrid vehicles”, “global warming”, “Kohleausstieg”…, in order to set up and maintain an uninterrupted data collection, lasting a viable amount of time. The participants of the research lab can split up in smaller groups for this task; each group can take care of a different source: real-time tweets collection, facebook discussions, newspapers, NGO/government websites, etc.
    3. Evaluate data analysis techniques for the identified types of data sources with the aim to transform the various data sources into a CCD (conceptual policy model). For example, consider descriptive statistics, also including network analysis metrics (modularity, degree distribution...), perform community detection at t=0, topic discovery (diagram frequencies, …); automatically obtain characterizations for communities and individuals (community 0 discusses about “health issues”, community 1 about “ethical considerations”, community 2 about “animal rights”...), individual N has a low in-degree but a very high out-degree, etc.
    4. Create a CCD from the collected information and acquired knowledge, covering as many actors and aspects as possible. Document the applied process and methodology for data analysis and transformation for this specific case. Students involved in the previous step would ideally also participate in this one.
    5. Optional depending on size of team, workload and skills: Develop a simulation model (otherwise this will be offered in follow-up works (e.g. master theses)).
    6. Document the research intern's results in the project's handbook (to be written in English).

    Required Qualification

    • Have good knowledge on analytical and empirical data analysis methods and tools
    • Have knowledge in conceptual modelling techniques
    • Have basic knowledge in programming code
    • Soft skills required:
      • Competences related to project management
      • Good analytical skills
      • Excellent documentation skills
      • Good English skills
      • Good communication and presentation skills

    Your benefits

    • Acquire knowledge and experience in engineering data for policy analysis and simulation
    • Acquire knowledge and hands-on experience in conceptual modelling
    • Become familiar with simulation methods
    • Experience project management and specific project roles in a team, including leadership
    • Experience documentation of project results and testing of models

    Sources

    [1] Scherer, Sabrina; Wimmer, Maria A.; Lotzmann, Ulf; Moss, Scott; Pinotti, Daniele (2015): An evidence-based and conceptual model-driven approach for agent-based policy modelling. In: Journal of Artificial Societies and Social Simulation. volume 18 (3), Paper 14. http://jasss.soc.surrey.ac.uk/18/3/14.html  

    [2] Scherer, Sabrina; Wimmer, Maria A.; Markisic, Suvad (2013): Bridging narrative scenario texts and formal policy modeling through conceptual policy modeling. In: Artificial Intelligence and Law. Volume 21 (4), pp 455-484 https://link.springer.com/article/10.1007/s10506-013-9142-2

    [3] Lotzmann, Ulf; Wimmer, Maria A. (2013): Evidence Traces for Multi-agent Declarative Rule-based Policy Simulation. In: Proceedings of the 17th IEEE/ACM International Symposium on Distributed Simulation and Real Time Applications (DS-RT 2013). pp. 115-122. https://ieeexplore.ieee.org/abstract/document/6690501

    [4] Lotzmann, Ulf; Wimmer, Maria A. (2013): Traceability in Evidence-based Policy Simulation. In: Rekdalsbakken, Webjorn; Bye, Robin T.; Zhang, Houxiang: 27th European Conference on Modelling and Simulation, ECMS 2013. Digitaldruck Pirrot GmbH: Dudweiler. pp. 696-702. http://www.scs-europe.net/dlib/2013/ecms13papers/pm_ECMS2013_0154.pdf

    Base textbooks

    Janssen, Marijn, Wimmer, Maria A. and Deljoo, Ameneh (2015): Policy Practice and Digital Science. Integrating Complex Systems, Social Simulation and Public Administration in Policy Research. Springer International Publishing

    Gilbert, Nigel and Troitzsch, Klaus G. (2005): Simulation for the Social Scientist. 2nd edition. Open University Press, McGraw-Hill, Maidenhead

    Newman, Mark (2010): Networks: An Introduction. Oxford University Press, Inc.

    Lecturer