Computer Engineering, University of California
The database model presented in this paper is suitable for application in which queries may require non-crisp references to certain attributes. The data item (attribute) values may be crisp or fuzzy. For instance, such adjectives as 'high' or 'normal' may be attribute values for the attribute blood pressure. A disease or a condition can be described by a number of symptoms which may be crisp alphanumeric values or fuzzy terms such as 'high' or 'normal'. A query into this database can retrieve diseases which have 'similar' symptoms. The similarity or 'indistinguishabiIity' is a measure defined by the database user on the relations that describe a family of diseases. database system in conjunction with a rule base can provide the framework for a medical consultation system.