A dermatologic decision support system that processes lexical knowledge from standard textbooks

We present a decision support system that processes lexical fuzzy knowledge from standard pathology textbooks. The user can choose among four methods: full text and hypertext approaches (classic lexical retrieval methods) and two possibilistic, knowledge-based approaches (the fuzzy logic computation and an adapted evidence combination scheme based upon the theory of belief functions). All these methods, including their algorithms, are presented in this paper. The effectiveness of these approaches is demonstrated by examples from dermatopathology. At the top of the lists of differential diagnoses generated by the system the degree of conformity between the approaches is high.