Victim-Centered Approach for Detection of Human Trafficking Victims
Process Flow and Scheduling
The process flow of our model is described with the pseudocode below .
The process flow will continue until all migrants have interacted with the state agent , and each migrant agent m _ i will only interact once per simulation run .
Implementation
Implementation Description
We implemented the model in Net- Logo ( Wilensky , 1999 ), an open-source , agent-based modeling platform . All agent updates are executed by default in NetLogo by using asynchronous random activation order .
Verification and Validation
Model verification was a continuous process throughout model development . The primary way to verify that the model was running as expected was through observing how the visualization changed as we manipulated model parameter .
To help with this process , we implemented agent visual representations to clearly indicate where in the process they were . These representations are shown in Figure 4 .
Figure 5 shows a panel plot for one sample realization of our model . Each plot shows the emergent pattern of migrant cooperation strategies for the given time tick t in the model . We observe that the migrant quota was reached at time t = 51 , but clustered communities of cooperation persist well into tick 500 .
Following Axtell and Epstein ( 1994 ), we suggest we have achieved Level 1 validation — our model can represent stylized facts for how influence and social contagion might plausibly spread through a migrant population , but is not at a level of realism that we can achieve statistical validity . This level of validation , while not good
Figure 4 . Visualization legend key for the model . For each agent type of a migrant or a seeded trafficking victim , the key shows the initial visualization for their intended strategy and the possible new state ( s ).
156