No abstract

No abstract

No abstract

No abstract

Conventional experimental/knowledge records (usually tables, graphs and/or equations) are not suitable for an efficient uncertainty-reasoning because of the poor knowledge-acquisition methods used to develop them. A fuzzy knowledge base is a suitable framework for acquisition of vague, sparse and inconsistent knowledge. A revitalization of valuable records and re-used literature knowledge items is a retrospective application of knowledge engineering algorithms. This is an ad hoe process and a general theoretical background does not exist. Therefore this paper presents a detailed description of a case study (49 variables, 4000 statements). A query as an element of a man-machine dialogue is presented. No apriori knowledge of fuzzy mathematics is needed. NOMENCLATURE a, b, c, d = description of piecewise linear grade of membership (see Fig. 1) A, = one-dimensional fuzzy set, quantification of the jth variable in the ith conditional statement B, = one-dimensional fuzzy set, quantification of a dependent variable in the ith conditional rn = number of conditional statements n = number of variables (dimensionality of knowledge bases) Q = n dimensional fuzzy query Ui = universe of ith variables X, = ith variable Y, = numerical value of dependent variable in the ith experiment statement S(n, V, 7') = similarity of two n dimensional sets V and T w ( Q ) = set of conditional statements activated by query Q Z,,j = numerical value of the jth independent variable Xj in the ith experiment p r ( y ) = grade of membership of y in fuzzy set T