The relational model versus the extended entity relationship model: A comparison of representations developed by autonomous users.

摘要

The inadequacies of the three classical data models in their ability to directly and naturally capture complex relationships between objects have led to suggestions of semantic data models. However, there is little empirical evidence that the semantic models lead to better user performance than the classical data models in conceptual data representation or any other database task.;To extend our understanding of this issue, a laboratory study was conducted to compare autonomous user performance between the relational model, which is a classical data model, with the extended entity relationship model, which is a semantic data model, during a representation task. Autonomous users may be defined as users who develop, design, implement, and use application programs in either interactive or personal computing environments to support personal or a small group's information requirement for decision making. A representation task involves the use of a modeling technique to represent the description of an application into a conceptual data model. A conceptual data model shows the objects of interest (entities), their properties (attributes), and the relationships between the objects.;A pilot study with twenty subjects was conducted to assess the feasibility of the proposed research, to refine task description, to estimate training and task completion times, and to propose a preliminary grading scheme. The main study was conducted with forty two subjects. During each experimental session, a few subjects were trained in the conceptual modeling task using one of the two data models. They were then asked to prepare a conceptual data representation of an application.;The solutions prepared by the subjects were graded for correctness using a reliable grading scheme. It was found that the EER model led to significantly better user performance in modeling binary and ternary one-many-many relationships. There was some evidence that the relational model was better in modeling unary relationship, and that the EER model was better in modeling ternary many-many-many relationship. Subjects found it difficult to model ternary relationships using either model, though the performance using the EER model was better. In general, there was evidence that the EER model led to better user performance. No differences, however, were found between the perceived ease of use of the two models.