Treatment of Vague Information in the Development of a Risk Evaluation System — Application to Seismic Risk Analysis

The development of an expert system involves processing many different types of information; some specific and some vague. Experts do have judgment and heuristic rules based upon their broad knowledge and experience. This judgmental knowledge often involves linguistic descriptors such as “serious,” “possible,” “important,” and so on. The way experts handle such information and reach assessment is also ill-structured and impossible to model using conventional mathematical means. This paper describes the use of fuzzy sets theory to incorporate vague information in the formulation of rules and the querying process when perfect matching no longer exists and contradictory statements are present. Two techniques, fuzzy system identification and the vertex method, are discussed in the treatment of information to obtain the risk functions. The models will be illustrated by means of examples derived from a seismic risk study.