In this work, a Web based Expert System (ES) for the identification of genders of gram-negative glucose nonfermenting bacilli is presented. These microorganisms are considered important agents in nosocomial infections, but its identification is a very complex process. The expert system’s knowledge base is conformed by two types of rules: primary rules, generated with the decision tree induction algorithm C4.5, but with some modifications to make one first classification in small groups of genders; and complementary rules, to characterize the gender. In order to handle the uncertainty, the certainty factor scheme was used. Tests made with isolated bacteria of different origin, show that the system allows a reliable characterization of the genders of a form simplified.
|