A Comparison of Uncertainty Calculi in an Expert System for Information Retrieval

This paper describes the f i r s t stages of an experimental i nves t iga t ion into the e f fec ts of using var ious c a l c u l i to propagate uncer ta in ty in an i n te rac t i ve expert system for in format ion r e t r i e v a l . We in te rp re t uncer ta in ty values as p a r t i a l t ru ths rather than p r o b a b i l i t i e s , and draw upon the mathematics of mul t i -va lued log ic in developing our ana lys is . We conclude that spec i f i ca t i on of an uncer ta inty calculus is a subt le problem that in te rac ts in several ways wi th the scheme used to represent the expert knowledge.