What's in a Relationship: on distinguishing property holding and object binding

This paper proposes improvements to the underlying data- or meta-metamodels used to represent information systems and IS development methods. It concentrates on the modelling of relationships, with particular reference to CASE and methodology engineering. Existing models are examined and some problems are identified arising from the confusion of the connection aspect of relationships with their property-holding aspect. The separation of attributes of relationships from information about what objects they link is proposed as a solution, and a concept of binding is proposed to model the linking information. The improved model is tested theoretically, semantically, and in practice on three metamodels