Inference method for fuzzy quantified and truth qualified natural languagepropositions

This paper proposes an inference method for fuzzy quantified and truth qual ified natural language propositions. For example, for "Most tall men are he avy is true," a modified proposition "Many more or less tall men are heavy is true," can be derived by inference. For the quantifier "more or less," t he fuzzy quantifier "many" can be derived analytically. Three types of fuzz y quantifiers (the monotonically nonincreasing type such as "few," the mono tonically nondecreasing type such as "most," and the single-peaked type suc h as "several"), as well as the monotonic and injection type truth quantifi ed qualifier such as true and false, are considered. For a proposition cont aining those quantifiers/qualifiers, it is shown that the fuzzy quantifier can be derived analytically by the fuzzy inference, for the modification to quantify the fuzzy subject part. (C) 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(2): 22-43, 2000.