The Problems of Data Modeling in Software Practice

This paper presents, from the author's perspective , the problems that occur in practice during data modelling. The author's experiences are a result of a considerable number of projects which he carried out in the framework of his consultancy role at USU Softwarehaus in MM oglingen (Ger-many). These projects concerned the following themes: Corporate Datamodelling Comparing Datamodels Project (Application)-related Data modelling. In all cases, E/R-notation was the chosen representation-form. From these experiences , the author formed an impression of the problems that occur in practice when deening a data model. These problems have, however, also led to the author's increased interest in knowledge representation , in turn leading to his usage of KR-methods in practice. This has shown itself to be quite eeective. Sections 2 and 3 brieey illustrate the recommendations and the experiences arising from their usage in projects. 1 Datamodelling in Practice-the Problems Datamodelling was still up until recently the buz-zword with which one believed to be able to solve the software crisis. CASE products concentrated on this area, meta-databases were created using a data-modelling process (E/R), and large companies invested millions in order to acquire a corporate data model. Although this trend has subsided a little , the theme in general is still of current interest. What Chen already recognised as an important ben-eet when presenting the E/R-Model, is today still seen as a key eeect of a data model: the representation provides a standard communications basis with which understanding between DP and users is more easily accomplished. This however, unfortunately seems to hold just for small data models. For larger areas of attention, the methodology starts to become ineeective, and no longer provides the overview required. Apparently, there are just a few 'gurus' who are able to create a complete complex data model. Often this data model quickly decreases in value, as soon as that person leaves the company. Director's ooces exist in which the corporate data model is hanging up behind glass-however, this is regrettably the only place in which the data model is noticed or paid heed to. The following problems, among others, have been recognised: 1.1 Low Expressivness of a Data Model in E/R-Form During the analysis phase, many of the organisa-tion's interdependencies and processes are identiied. These are subsequently, to use the relativly inadequate language of the E/R-Model, abstracted and generalised. This often requires a change in terminology ; in other words …