We use the term ontology here to indicate the model of the categories within the universe of discourse, plus sufficient information about those categories to allow them to be classified automatically. We take ontologies to be language independent, using the broader term terminology for an ontology linked to linguistic information. We also assume the ontology to be separate from more general inference systems dealing with time, causality and so on.
Despite the philosophical connotations of the word ontology, our work in modelling an ontology for computer applications has required a practical and pragmatic approach. Although agreement on a high level schema is a prerequisite for effective co-operation between different groups of modellers, it is not practically possible either to build, or construct an implementation to support, a philosophically perfect model. What follows, therefore, is not claimed as an ultimate or perfect solution. It is, rather, the current stage in a series of successive approximations to practical tools which will:
- Fit closely enough with accepted medical usage to enable intuitive user interaction, especially in applications for clinical data entry or the retrieval and classification of medical concepts
- Be re-usable, i.e. be sufficiently general to be used by a range of different medical applications
- Support practical collaboration between different groups, being easily understood, taught, and used