Toward software engineering for multimedia information computation based on fuzzy objects

Abstract

Describes the possibility of developing a software method that integrates fuzzy logic (linguistic expression) and ordinary crisp computation in an object-oriented framework. A few examples are carefully examined to reveal the essential characteristics regarding: meaning of fuzzy rules from the object-oriented view point; and formal description of integrated crisp and linguistic computation. Assuming that a human has the ability to recognize, understand, and process digital and linguistic data, it can generate output in the form of linguistic expression. The paper describes the following: (1) A possibility of integrating the crisp algorithm (the ordinary programming) and linguistic expressions. For example, the fuzzy graph concept is naturally introduced and expressed by our new notation. (2) Unavoidable ambiguity (multiplicity) in recognizing multimedia information at the linguistic level. This explains why and how multiple fuzzy rules must be evaluated every time while an ordinary system requires one rule applied and evaluated at a time. (3) A new inheritance called type fusion of the fuzzy-object oriented formulation. Because of (1), the linguistically identified objects need to evaluate their own methods and methods of other object types together. In this way, the new object class inheritance is defined.