Title :
Weight structures for approximate reasoning with weighted expressions
Author_Institution :
Dept. of Comput. Sci., Dortmund Univ., Germany
Abstract :
One method of constructing an `approximate reasoning´ system is to use a `classical´ system of many-valued logic and attach to each logical expression a `weight´ which assesses the validity of this expression. Several such systems have been described in the literature, with varying interpretations concerning structure and semantics of weights. In this paper, a `canonical´ principle for defining the fundamental relations model and semantic consequence for logics with weighted expressions is described, which not only allows a large variety of truth-value and weight structures, but furthermore allows to transfer the results of `classical´ model theory to the resulting logics in a natural way
Keywords :
fuzzy logic; inference mechanisms; multivalued logic; uncertainty handling; approximate reasoning; fundamental relations model; many-valued logic; semantic consequence; weight structures; weighted expressions; Computer science; Fuzzy logic; Fuzzy sets; Joining processes; Lattices; Logic design; Multivalued logic; Uncertainty;
Conference_Titel :
Multiple-Valued Logic, 1996. Proceedings., 26th International Symposium on
Conference_Location :
Santiago de Compostela
Print_ISBN :
0-8186-7392-3
DOI :
10.1109/ISMVL.1996.508373