The problem of labels or names

In principle, the GALEN Common Reference Model is language independent and would behave the same way regardless of the names or labels given to its concepts. However, it is impossible to discuss the model without using a set of labels or knowledge names (introduced by the keywords name , newSub and newAttribute in GRAIL) The behaviour of the model is always, at best, an approximation of our conceptual structures, and so the names or labels are always approximate.

In discussing computer models, it is crucial to separate Does this entity’s behaviour in the model represent a useful concept? from Does it behave in the manner suggested by its label?. The labels given to symbols in models never fit perfectly and, to understand a concept, one can not rely only on the apparent usual interpretation of its label. It is often hard to find different and appropriate labels for new concepts when they differ only a little in intension from concepts already existing in the model. This is especially true of the various possible formal definitions of abstract concepts, such as disease or phenomenon. Labels must be chosen to make models usable, and any subsequent arguments should be clear when they are about the function of the model and when about an interpretation implied by a particular set of labels which can be easily changed.

In general, labels or knowledge names have been chosen to try to reduce ambiguity. In a multilingual project, the primary names are usually derived from English, but wherever there has been confusion, the names have been expanded to try to reduce ambiguity. In general, the approach has been that it is better that users and developers recognise that they do not understand a concept in the model from its name than that they assume incorrectly that they understand it from its name.

Making the impossible very difficult, ©, All rights reserved