KEYNOTE 4: Classification and Science
November 9th, 09:00 - 10:00
Classifying phenomena is deeply intertwined with cognition and human information processing. Therefore, identifying classes is a central aspect of information technology (IT). Choosing a “good” set of classes is both theoretically and practically important. Two cognitive principles underlie the cognitive approach to classification. First, classes encapsulate inferences about the properties of their instances – in other words, knowing a category can “tell” us more about an instance that required to identify the category it belongs to. Second, collections of classes should provide economy of storage. This leads to a view of classes as carriers of domain knowledge in the form of inferences about situations, which is more than “containers” for information.
We discuss how this view can be used to model scientific theories. We explain how the principles can be used to guide the choice of collections of classes. We show how the approach can be used in scientific discourse by applying it to one of the most well-known areas of physics – the electromagnetic equations as developed originally by Maxwell. The example shows how the classification based approach can be generally applied to scientific problems and that it has two advantages. First, it can provide a simpler and more informative account of the sample phenomena. Second, the classification principles can lead to questions to be asked to help resolve differences between observations and predictions. This means that the resolution of problems can be framed in terms of changes to classification structures, and to principles suggesting how such changes might occur.