ESSA Special Interest Group Qual2Rule – Using qualitative data to inform behavioral rules
Many academics consider qualitative evidence (e.g. texts gained from transcribing someone talking or observations of people) and quantitative evidence to be incommensurable. However agent-based simulations are a possible vehicle for bridging this gap. Narrative textual evidence often gives clues as to the in-context behavior of individuals and is thus a natural source for behaviors to inform the specification of corresponding agent behavior within simulations. They will not give a complete picture of this, but they will provide some of “menu” of behaviors that people use. During this session we hope to further understanding of how to do this better. It is open to all approaches that seek to move from qualitative evidence towards a simulation in a systematic way. This includes, but is not limited to:
- Approaches based in Grounded Theory.
- Tools for facilitating such a process.
- Participatory processes that result in a simulation.
- Frameworks for aiding the analysis of text into rules.
- Elicitation techniques that would aid the capture of information in an appropriate structure.
- Models and ideas from psychology to aid in the above process.
- Insights and tools from Natural Language Processing that may help this process.
- Agent architectures that will facilitate the programming of agents from such analyses.
- Philosophical or Sociological critiques of this project, pointing out assumptions and dangers.
- Examples of where this approach has been tried.