The AXIOM approach for probabilistic and causal modeling with expert elicited inputs

Expert informants can be used as the principal information source in the modeling of socio-techno-economic systems or problems to support planning, foresight and decision-making. Such modeling is theory-driven, grounded in expert judgment and understanding, and can be contrasted with data-driven modeling approaches. Several families of approaches exist to enable expert elicited systems modeling with varying input information requirements and analytical ambitions.

This paper proposes a novel modeling language and computational process, which combines aspects from various other approaches in an attempt to create a flexible and practical systems modeling approach based on expert elicitation. It is intended to have high fitness in modeling of systems that lack statistical data and exhibit low quantifiability of important system characteristics. AXIOM is positioned against Bayesian networks, cross-impact analysis, structural analysis, and morphological analysis. The modeling language and computational process are illustrated with a small example model. A software implementation is also presented.