Formal Techniques of Data Structure Design in Modern Database Systems

Abstract

An overview of current status in the area of formal technique of object-oriented database design is described in the article. It is discussed here, why classical relational design techniques as normalization, decomposition and synthesis are not able to be easy used in modern databases. The article introduces various proposals of object normal forms and it brings also own authors’ approach of object normalization. Based on our experience with agricultural I.S. development, it is reasonable to assume that the importance of object-oriented and post-relational databases will grow in the near future, because there are now many possible applications, where object database shows practically its advantages. Large amount of complex data structures and their variability during system operation is the common attribute of these applications, which cause big problems to relational databases. These systems can work with more then hundred or thousand various mutually associated and composed data types represented by object classes as it in the object-oriented programming languages. The queries over these objects require moreover a high rate of common polymorphism. (In these systems, for example, we need to write down the queries over sets containing elements of various types. At the same time we expect that while adding of a new query it will not overwrite the already executed queries.) Good example of these systems is data warehouses, which are specific by their long-time collections of large amounts with various kinds of data. These systems go about farm company management, various regional management systems, and information systems containing ecological information, agricultural information and decision support in business