A proposed fuzzy RDBMS and its test results on an osteoporosis patient database



 The new trend in database design is to store both crisp and inexact data, and have the possibility of fuzzy querying in both crisp and fuzzy databases. Traditional databases use only crisp data and they are only capable of manipulating them using Boolean logic, while in a reasonable number of situations we are faced with uncertain, approximate and vague data. In those situations, it may not be reasonable to transform these data into crisp forms. Fuzzy databases, on the other hand, are capable of storing and retrieving both, crisp and non-crisp data and to manipulate them using fuzzy logic. This paper proposes a possibility-based fuzzy relational database management system. Mechanisms to design and implement the database are shown and query processing methods are described. Fuzzy queries are first translated into SQL and then passed to the MS-SQL server to process them. The result is finally fuzzified. A user can express his (her) request using linguistic terms and can choose different retrieval threshold values. The system is then used to store information about patients whose bone density is measured for diagnosis of osteoporosis. It is shown to be more flexible and has produced more accurate results compared to a crisp database