Unformatted text preview: above, we pay particular attention to inheritance, overriding and late binding, and completeness. The other features are captured as a by-product. But before we proceed with the development of ORLog, we must also develop a data model that can be used as an underlying conceptual tool for ORLog schema design and conceptual modeling. To that end, we develop a conceptual model, called the OR model, that combines features from SDMs and object-oriented programming, giving rise to the notion of abstract data models. Finally, we also present a prototype implementation of ORLog into Coral deductive database system. To this end, we rst develop an encoding technique to translate ORLog to Coral and an inheritance reduction technique to be able to implement inheritance in a purely deductive system. 1.3 Organization of this Thesis
We organize the presentation of this thesis as follows. We present a complete theme in a single chapter. Since we address several unique issues that have distinct properties and features, we delegate the discussion on their background and the related research in that area in the corresponding chapters. Hence we do not include a separate chapter on related works and motivations in this thesis. Whenever appropriate, we include our comments at the end of a particular chapter. Keeping the above in mind, in Chapter 2, we develop and present our abstract data model called the Object Relationship model, or the OR model. Then in Chapter 3 we present our language ORLog that embodies the semantics of all the seven mandatory 3 features, and behavioral inheritance in particular. ORLog is a logical formalization of the OR abstract data model. We discuss a translation technique of ORLog to a rst order language in Chapter 4. In this chapter, we also present an inheritance reduction technique based on completion that together with rst-order translation enables us to implement ORLog in an existing deductive database system, such as Coral, as an object-oriented front-end. Finally we also report and discuss a prototype implementation in Chapter 5 based on the reduction technique presented in Chapter 4. In Chapter 6 we present a critical review of contemporary research in logical object-oriented paradigm in view of ORLog and discuss issues related to ORLog implementation. We then summarize and give our conclusion in Chapter 7 and discuss future research. 1.4 Credits
Di erent parts of the work reported in this thesis have been carried out in collaboration with various people. The purpose of this section is to acknowledge their contributions and attribute credits. The conceptual OR data model described in Chapter 2 is a joint work with Dr. Lakshmanan and Mona Soliman. The logical foundations of ORLog, its syntax, declarative, xpoint and proof-theoretic semantics presented in Chapter 3, and the reduction of ORLog inheritance to deduction discussed in Chapter 4 as well as the results in these chapters were established in collaboration with Dr. Lakshmanan. Finally, implementation of ORLog in Coral was done jointly with Dr. Lakshmanan and Ianina Orenman. 4 Chapter 2 OR Model: An Abstract Data Model
In this chapter we shall present our Object Relationship (OR) model. We will rst outline the broad objectives of this new data model. Then we will describe various building blocks, or semantic constructs, we use in our model. While some of the constructs are adapted from other semantic data models, several new ones are introduced to increase the modeling capability of the OR model. We then discuss global restrictions on the building blocks in order to be able to use these constructs in a coherent manner for database schema design. We also propose a notion of admissible databases which captures the intended semantics of a meaningful database by imposing further restrictions on the semantic behavior of the model. We will also identify its strengths and weaknesses by comparing it with contemporary data models as we describe our model. But before we proceed, we present a survey of existing data models for object-oriented databases. 2.1 Related Research and Motivation
Programming with taxonomically organized data which are encapsulated with operations has led to object-oriented programming. The object-oriented programming paradigm is not only able to handle and manage complex and hierarchically structured data, it is also able to express the dynamic aspects of data with the introduction of the concept of behavior and methods. The notion of inheritance in object-oriented 5 programming is intimately related to re-usability and sharing. The notion of encapsulation brings modularity and implementation independence to this paradigm. It is believed that combining object-oriented concepts with databases will result in clear bene ts 77]. There is currently signi cant interest in object-oriented database systems. Much of the research on OODBs was sparked by the need to provide adequate database support for next generation database/knowledge-base applications such as engineering design, software development, VLSI design, etc. In general, such applications necessitate features such as complex ob...
View Full Document