?
Subject-formal Methods Based on Cognitive Maps and the Problem of Risk Due to the Human Factor
The chapter gives a brief survey of contemporary subject-formal methods based on cognitive maps, which are used to solve control problems for ill-structured objects and situations. These methods combine mental processing of initial knowledge of a situation with formal methods of cognitive maps processing (simulation or other formal methods).
The essence of the problem of risks due to the human factor while applying subject-formal methods is explained. The essence is that due to inevitable and substantial humans’ participation in solving practical problems for complex and ill-structured objects and situations (at least, for formalization of primary representations) formal methods basically cannot provide validity of received solutions, and therefore these methods are risky.
The chapter describes a number of inter-disciplinary models that can be used to analyze, explain, and forecast risks due to the human factor in the whole lifecycle of subject-formal methods: the distorting effect model, the stereotypes effect model, the ambient intelligence effect model, and the generalized model of expert-analyst.
A number of risk factors that arise while simulating the situation dynamics on the base of cognitive maps is disclosed. Amongst them are the first-kind risk factors coming from decision-making personnel (factors of direct action); the second-kind risk factors coming from theorists and computer-aided technology designers (factors of indirect action); and the hybrid risk-factors.
In order to reduce the risks due to the human factor a number of heuristic criteria of formalization validity is proposed and proven. These criteria can be applied to early disclose direct semantic errors and formalization risks.
The functional approach to verification of different types of models based on cognitive maps, with fuzzy semantics is proposed.
Some of the discovered risks are demonstrated with the use of cognitive maps found in scientific literature as well as maps that have been used to solve real control problems.
The main challenges and directions of further research are presented.