Victim-Centered Approach for Detection of Human Trafficking Victims
Figure 2 . Model screenshot for a sample realization .
Agents
There are two agent types in the model — migrant agents and one state agent .
Migrant Agent
Migrant agents are the primary agent objects of the model and make up the population of individuals , whom we assume for the context of this research as a migrant population congregated awaiting to be processed by the border authorities of the state . We denote a migrant as m i
M , where M
is the agent set of all migrants in the population .
Migrant Agent Instance Variables Each migrant m i is heterogeneous and characterized by the instance variables shown in Table 1 . Instance variables for s i
,
k i
, and v i were meant to be calibrated to real-world data if possible , with s i representing whether the migrant is one of our seeded trafficking victims . This may be directly calibrated to existing data . For example , Germany reported that 5.5 % of trafficking victims were self-identified in 2015 ; thus , we may seed a similar proportion of the migrant population as victims that need to be identified . The variable for skill level k i may be “ high ” ( represents the migrant having a high school degree ) or “ low ” ( as not having a degree ), and v _ i represents the country of origin and the circumstances of that country . Using Figure 1 as an example , we assume that migrants arriving from Syria , Afghanistan , and Iraq are fleeing , while migrants from all other countries of origin are seeking .
Cooperation intention y i is randomly assigned to each migrant agent during instantiation as per a user-specified parameter for proportion of initial cooperators — agents are preassigned with an intention to cooperate or deceive .
152