Simulations

Debates and Positions are static objects: they describe arguments and an agent’s belief system at one point in time. Dynamical aspects of debates and belief systems can be studied in Simulations.

In agent-based models built on the theory of dialectical structures, the simulated world consists of a sentence pool and arguments made from these sentences. Agents have a multi-dimensional belief system in terms of Positions. The Simulations progress by introducing arguments, possibly according to an argumentation strategy assigned to agents. Arguments can be introduced in a random fashion to debates. Those have minimal requirements and can be used even in Simulations that do not contain any agents. Purposeful argumentation strategies select premises and conclusions in light of agents’ belief systems. For these arguments, at least two agents are drawn from the population. After argument introduction, the entire population responds by checking their belief system in light of the new argument and revise their beliefs if necessary. This process continues until a termination condition is reached. If requested by the user, the sentence pool is also occassionaly extended. The simulation terminates when the desired inferential density is reached or the desired number of arguments has been introduced.

• Setting up a population
  • Populations are lists of positions
  • Argumentation strategies
• Setting up a simulation
  • Examples
    • A minimal example
    • A more realistic example
  • Simulation types
    • Iterative argument introductions
    • Pre-compiled argument maps
• Running a simulation
• Experiments: Simulations in parallel
• Evaluation
  • Setting up an Evaluation object
  • Viewing results
  • A minimal example
  • Adding data to an Evaluation object
    • Shortcut functions
    • Measures that only analyse debate stages
    • Measures that only analyse positions
    • Measures that rely on clustering