Methodological Principles of Uncertainty in Inductive Modelling: A New Perspective

摘要: It is argued that the concept of uncertainty plays a fundamental role in inductive (data-driven) systems modelling. In particular, it essential for dealing with two broad classes problems are to modelling: involving ampliative reasoning (reasoning which conclusions not entailed within given premises) and simplification. These problem closely connected principles maximum minimum uncertainty. When models conceptualized terms probability theory, these become well established entropy. However, when more general framework Dempster-Shafer theory evidence employed, four different types emerge. Well justified measures now available described paper. The meaning captured by suggestive names “nonspecificity”, “fuzziness,” “dissonance,” “confusion.” Since multidimensional entity lead optimization multiple objective criteria.