A Synthetic View of Approximate Reasoning Techniques

Abs t rac t : This paper presents a review of d i f f e r e n t approximate reasoning techniques which have been proposed for dea l ing w i t h uncer ta in or imprecise knowledge ,espec i a l ly in expert systems based on product i o n ru le methodology. Theore t i ca l approaches such as Bayesian i n fe rence ,Sha fe r ' s b e l i e f theory or Zadeh's p o s s i b i l i t y theory as wel l as more emp i r i ca l proposals such as the ones used in MYCIN or in PROSPECTOR, are considered. The p resen ta t ion is focused on two basic in ference schemes : the deduct ive i n f e rence and the combinat ion of several uncer ta in or imprecise evidences r e l a t i v e to a same mat te r . Several kinds of unce r ta in t y are taken i n t o account in the models which are descr ibed in the paper : d i f f e r e n t degrees of c e r t a i n t y or of t r u t h may be associated w i t h the observed or produced fac ts or w i t h the " i f . . , t h e n . . . " r u l e s ; moreover the statements of fac ts or of ru les may be imprecise or fuzzy and the values of the degrees of c e r t a i n t y which are used may be only approximately known. An extensive b i b l i og raphy , to which i t i s r e f e r red in the t e x t , i s appended.