Learning and generalization of characteristic descriptions: evaluation criteria and comparative review of selected methods

摘要: Some recent work in area of learning structural descriptions from examples is reviewed light the need many diverse disciplines for programs which can perform conceptual data analysis. Such describe complex interms or logical, functional, and causal relationships cannot bediscovered using traditional analysis techniques. Various important aspects problem are examined criteria evaluating current presented. Methods published by Buchanan, et. al. [1-3, 20], Hayes-Roth [6-91, Vere [22-25], analyzed according tothese compared to a method developed authors. Finally some goals suggested future research.