RDF: A Framework for Specifying Ad-hoc Rules on ORM Schemas, Jon (J. I.) McCormackORM-1 (First International Conference on Object-Role Modelling) Australia, 1994.
The paper proposed the Rule Definition Framework (RDF) as a mechanism for adding ad-hoc rule specifications to ORM (Object-Role Modeling) conceptual schemas. RDF provides a flexible semantic layer on top of ORM schemas so that ad-hoc business or domain rules—beyond what classical ORM constraints express—could be captured and processed in a unified framework
Many CASE tools for information systems engineering can input a conceptual data model of an application and map this to a logical data model for implementation. Typically this involves mapping an ER (Entity-Relationship) conceptual schema to a relational database schema. Since the graphic notation of ER, or the mapping algorithm itself, fails to capture many constraints and derivation rules, these additional features must be coded up manually. Object-Role Modelling (ORM) provides a simpler and richer notation, enabling most of these additional features to be catered for in the mapping. The most well known version of ORM is NIAM, and a number of CASE tools now support this method. Recently, an extended ORM language called FORML has been developed which is even more expressive, and a complete mapping algorithm has been developed and automated. This paper provides an overview of the mapping algorithm and the use of role-graphs for automation.
For a database application, conceptual design methods such as fact-oriented modelling and entity-relationship modelling are commonly used to specify a conceptual schema, which may then be mapped to a structure in a chosen data model (e.g. a relational database schema). Since conceptual data models support a rich variety of constraints, and these constraints may impact on one another, the task of ensuring mat the constraints expressed in a conceptual schema are consistent is non-trivial. Moreover, because different constraint patterns may be equivalent, some optimization may be needed to select the best constraint pattern for explicit assertion. With reference to conceptual schemas expressed in FOrML (an enhanced version of NIAM) mis paper discusses metarules for strong satisfiability and constraint preference, and outlines an efficient algorithm for validating four main types of constraints. Complexity analyses and benchmarks of the implemented algorithm are included.
