Getting Ready for Session #16 (October 28, 1996)
In our consideration of representation of information or knowledge, we have been concerned with issues of what sorts of information/knowledge might be important (objects, attributes and relations) and of how that information/knowledge might appear (format & values). That is, if we decide that date needs to be represented within our system, then at some point we need to decide whether year is enough or whether we need month, day, hour, minute, or second. We also need to be concerned with whether May 9, 1977 at 11:59 pm, 5.9.77-23:59, or some other format will be better for our particular purposes.
While there are other purposes and uses for a classification, we might focus our attention at least for the moment, on the need for "authority" when assigning values to elements of information that have been selected for representation. For instance, if we have a data or knowledge base that has something to do with work projects, one aspect of these work projects that is represented might be type of project. We might 1) let whomever is entering data designate a project type in any way that they wish (e.g., river, riverine, stream, creek, etc.) or 2) we might specify those project types that are acceptable within our system (e.g., river for all of the variations). For 1) we would need to indicate all of the kinds of project types that were of interest to us in order to retrieve all of the records or frames that were of concern to us. For 2) we need only enter the standard project type. In our discussions of knowledge representation strategies, the issues of "classing" and specifying hierarchies of classes and subclasses came up, particularly in connection with frames. We will be spending the next sessions considering classification strategies and their strengths and weaknesses.
During this introductory session we will begin to consider what we gain and what we loose as we order, categorize, classify, index, etc. In this light the reading lays a foundation.
Norman, D.A. (1993). Things that make us smart : defending human attributes in the age of the machine. Reading, MA: Addison-Wesley. [T14 .N67 1993] Ch. 7.
I predict (but what do I know) that you will have fun with this one. Norman moves us from the days of the "Wooten patent desk" to more modern "solutions" to classification and order problems. Whether we accept his judgments or opinions or not, I think that he will challenge your thinking, if you let him!
Ask yourself as you read this selection: