Echoing Juan Carlos’ post on applications of digital humanities, I’ve seen exciting opportunities to apply digital methods in social science as well, especially in my field, urban studies. Besides data visualization, digitization, spatial analysis and representation, social network analysis, and text analysis covered in his blog, I also think agent-based modeling is gaining popularity.

An agent-based model (ABM) is a computational model that simulates the behaviors and interactions of different agents in a dynamic system. Its theoretical foundation is rooted in game theory, cellular automata, complex systems, and others. One interesting application I read recently is to use ABMs to simulate urban food deserts, which are traditionally studied by surveys or case studies. For example, Auchincloss et al. (2011) built an ABM of income inequalities in dietary quality to study the role of economic segregation and test possible interventions. In their model, there are two types of agents: households and food stores. In each time step, a household chooses a store to shop for food. The criteria for selecting food include a store’s food price, distance to a store, the household’s habitual behavior, and the household’s preference for healthy food. Stores’ behaviors depend on demand and households’ ability to make new store choices. Store closures and openings are programmed accordingly. This ABM reveals that the segregation of high-income families and healthy food stores from low-income families and unhealthy food stores leads to income disparities in diet.

ABM is considered to be not only a simulation technique but also a mindset. The latter requires system thinking and examining individual constituent units of a system. The complexity of measuring parameters in social science adds to the challenges of building ABMs. Eric Bonabeau, one of the top experts in complex systems, argues that for human system modeling, results of AMBs should be interpreted at the qualitative level (Bonabeau, 2002). In other words, ABMs serve better as learning tools to understand social phenomena than as predictive tools. When involving human factors, he summarizes that it is best to use ABMs:

– When the interactions between the agents are complex, nonlinear, discontinuous, or discrete.

– When space is crucial and the agents’ positions are not fixed.

– When the population is heterogeneous, when each individual is (potentially) different.

– When the topology of the interactions is heterogeneous and complex.

– When the agents exhibit complex behavior, including learning and adaptation.

Relating to cultural heritage, I found that ABMs have been widely used in archaeology, philosophy, history, and more. Anyway, this post is just to add to Juan Carlos’ list of applications of digital humanities.

References:

Auchincloss, A. H., Riolo, R. L., Brown, D. G., Cook, J., & Diez Roux, A. V. (2011). An Agent-Based Model of Income Inequalities in Diet in the Context of Residential Segregation. American Journal of Preventive Medicine, 40(3), 303–311. https://doi.org/10.1016/j.amepre.2010.10.033

Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human systems. Proceedings of the National Academy of Sciences, 99(Supplement 3), 7280–7287. https://doi.org/10.1073/pnas.082080899