I highly recommend works and papers by Epstein and Axtell- computational modeling of social phenomenon, macro structures derived as a generative product  of micro motives/states. Features of recursion and contagion emphasized in their models.

Doran's work on social simulation will provide interesting material for the algorithmic treatment of agents' cognitive psychology.

At a spatial level and within an urban ecology, works by Batty. In particular, visualization features of his simulations.

https://www.researchgate.net/profile/Joshua_Epstein/publications

https://www.researchgate.net/profile/Robert_Axtell2/publications

https://www.researchgate.net/profile/James_Doran/publications

https://www.researchgate.net/profile/Michael_Batty2/publications

Hi Alicia,

      Hope you're doing well!

      I may not be the best person to answer this question, since I do not actively monitor the research literature on this topic.  My focus tends to be at the other end of the issue:  what methodology or approach has the potential to produce demonstratively valid multiscale social models?  Regarding the latter question, it seems to me that homotopy type theory (cf., http://homotopytypetheory.org/book/), when focused on social issues/models, has the greatest potential to produce, demonstrate &/or ground multiscale model validity.  

      One reason that HoTT (the standard abbreviation) has this potential resides in its expressiveness.  The nature of type theory supports the scale variation that you (rightly) seek.  This expressiveness reaches ever higher levels:  from objects to morphisms to functors to higher categories.  We have yet to see it effectively applied to social categories, but the capabilities, and therefore the potential, of HoTT is impressive.

     A second feature of HoTT is its ability to support inference based on equivalence.  This supports the integration of quantitative and qualitative models, and allows models to coevolve toward greater rigor.

      A third feature of HoTT is its support for higher induction (chapter 6), which allows social theories, models and generalizations to be mapped to specialized spaces and shapes.  This is a means of providing rigor for custom or idiosyncratic models and, ultimately, the relation between distinct but related social mechanisms.  As you can see, this is another aspect of HoTT expressiveness.

     A final example (for this response) of HoTT capabilities is that it supports 'propositions as types'.  This means that an ability to support proofs is built into the HoTT framework.  Of course, it is primarily designed to support mathematical proofs but, to the extent that social theories are expressed within the HoTT framework, social premises and theorems can be explored and tested as well.

      As I implied in my description of the first HoTT capability, we are some distance from demonstrating this potential.  However, in my view, HoTT features and capabilities are impressive and pertinent to the multiscale  integration and assessment of social models, and it is only a matter of time (and effort!) before this potential will be realized.

Regards,

David

Dr. Shian-Loong Bernard Lew, thank you for the references you provided. I have searched through the publications of the researchers you mentioned, and I have found some interesting content. Specifically, I found these two papers:

G. V. Bobashev, D. M. Goedecke, F. Yu, and J. M. Epstein, "A hybrid epidemic model: combining the advantages of agent-based and equationbased approaches, " in Proceedings of the 39th conference on Winter simulation (WSC'07). IEEE Press, Piscataway, NJ, USA, pp. 1532- 1537, 2007.

Epstein, J. M., Pankajakshan, R., & Hammond, R. A. (2011). Combining Computational Fluid Dynamics and Agent-Based Modeling: A New Approach to Evacuation Planning. PLoS ONE, 6(5), e20139. doi:10.1371/journal.pone.0020139

I wasn't familiar with either Doran or Batty, but their work is quite interesting. I expect to be taking a closer look at both of their research efforts in regards to other projects I'm working. 

Thanks again,

Alicia Ruvinsky

Hi, David.

Thanks so much for answering my question! Your feedback was very thought provoking. I have also been looking through your publications and found this one very interesting and relevant:

Social Theory and Agent Architectures: Prospective Issues in Rapid-Discovery Social Science

It has provided me with a framework with which to think about model integration (e.g., microtranslation, micro-situation).

There are a lot of factors to think about here, and I’m still trying to get a good grasp on what this problem looks like. The state of the art for model integration appears to be pretty straightforward aggregation of output (like what is done on projects like ICEWS). The aggregation is a little intelligent in that it’s Bayesian-based, but I'd say straightforward because it doesn’t go beyond combining output (i.e., conceptual and theoretical integration a la MESA).

More extensive model integration that combines models within execution (e.g., one or more models providing regular input to another model) would require an assessment of (1) the appropriateness of integrating certain models, and (2) the meaning of integration certain models. If model_A provides input to model_B, what is the data model_A is providing and what does mean? How should model_B make heads or tails of model_A’s data? This is very much where HoTT would fit. HoTT would provide a logical infrastructure by which to (1) define model_A’s data, and (2) understand what model_A’s data is (based on what model_A does) so that a reasoning mechanism can be appropriately applied. So, model_A can effectively communicate what it is doing, and model_B can effectively leverage model_A’s data to do work.

I am now in the process of looking deeper into the reference you sent and other sources on HoTT. I’m curious to know if a small experimental proof of concept can be designed using HoTT to specify model metadata and use this metadata to build a logical argument enabling model integration. This would be interesting.

I also hope that my intuition pans out that the MESA epistemological hierarchy will be relevant and useful in this space. I've attached a reference to a write-up of the work that we did for the Model Evaluation, Selection and Application (MESA) project.

Thanks again for taking part in this discussion, David.

All the best,

Alicia

Conference Paper Establishing Bounds of Responsible Operational Use of Social...

Dear Alicia,

You're most welcome.

All the best,

Bernard