EVIDENCE-BASED AGENT MODELING

Within the article we present a novel methodology of constructing agentbased models (ABM) on the basis of evidencebased modeling concepts of G.B. Kleiner. We redefine modeling as finding a computable mapping between two essentially mathematical object (enclosing and enclosed ABMs). We then introduce correct models space – a fourdimensional lattice, where the axes correspond to compression, evolution, emergence and convolution operators (defined over various types of sets via close linkage). Within our paradigm moves every move along a given lattice dimension represents adding a corresponding operator to the composition, while defining exact operators and their final composition (path on the lattice) yield a complete and formal set of model assumptions. On top of that we propose a stricter definition of emergence via dynamic systems attractors, a compact definition of natural language, as well as alternative interpretation and potential generalization of the probabilistic modeling approach.