A Graph-Oriented Object Model for Database End-User Interfaces

Abstract

The current database research trend is towards systems which can deal with advanced data applications that go beyond the data standard "enterprise" of "office" database application. This trend is reflected in the research on extension architectures and object-oriented databases. Along with this trend, the need for better and easier-to-use database and end-user interfaces has been stressed. Therefore, we propose a graph-based data model which shares many features with existing data models, but which better facilitates the rigorous study of graphical database end-user interfaces. Graphs have been an integral part of the database design process ever since the introduction of semantic data models. Their usage in data manipulation languages, however, is far more sparse. To deal with data manipulation,typically, schemes in semantic data models are transformed into a conceptual data model such as the relational model. The required database language features then become those of the conceptual model. Object-oriented data models on the other hand, often offer computational complete, non-graphical data languages, usually in the style of object-oriented programming languages such as Smalltalk. Due to their expressiveness, however, these languages do not lend themselves easily as high-level data languages. The first graphical database end-user interfaces were developed for the relational model (for example Zloof's Query-By-Example (QBE)). The earliest graphical database end-user interfaces for semantic models were associated with the Entity-Relationship model. Subsequently graphical interfaces were developed for more complex semantic and object-oriented database models. These interfaces use graphs as their central tool, but as far as data languages, they are usually limited in expressive power. Graph-oriented end-user interfaces have also been developed for recursive data objects and queries. In an earlier publication we introduced the Graph-Oriented Object Database Model(GOOD). This model is built around a single mathematical tool, namely graphs, to both model and manipulate databases. We believe that this is an important step in the direction of rigorously studying and developing database end-user interfaces. In that publication we limited ourselves to describing a simple yet powerful transformation language and discussing its expressiveness. in this paper, we further develop and investigate GOOD. We show that it has many features generally present in existing semantic, object-oriented and deductive database models. Specifically, we demonstrate how the GOOD model is suitable for graphically describing, querying, browsing, restructuring and updating databases, and hence is ideally suited for the study and development of graphically-oriented database end-user interfaces. To demonstrate why GOOD is useful for advanced data applications, we describe how it can be seen as an object-oriented data model. In Section 2 we define the basic GOOD model. In Section 3, we discuss querying, browsing, restructuring and updating, and show that they all can be expressed naturally in a uniform, graphically-oriented and user-friendly manner. We also show how to use GOOD to manipulate and query database schemes. In Section 4, we show how to adapt the GOOD model to incorporate the features of object-oriented database systems.