This class is an experiment.  In many fields, such as architecture and the arts, studio instruction is used to provide a more "hands-on" way of learning.  Instead of several short meetings a week, which are optimized for lecture courses, we set aside one large piece of time so that issues can be explored in more depth and significant creative work can be accomplished.  I think studio instruction is a good model for programming courses and for other courses that require a lot of creative thought.  Working out how to represent knowledge seems to be one of those areas.

As far as I know, this is the first time someone has tried to teach knowledge representation as a studio course.  KR makes great sense for this approach, because tackling representation problems requires deep conceptual analysis and working through examples, ideally with computational support.  And that is where the second part of the experiment comes in.  One can run versions of sophisticated knowledge-based systems on everyday computing equipment.  That means by using notebooks (or Tablet PC's, depending on availability), we can use tools that help build and test our representations on the spot.  This should provide a kind of reality check that enables us to explore more deeply. 

Here is a model of how our sessions will work, in steady-state:

  1. I'll start with a brief introduction to the representation problem we're going to tackle. 
  2. We'll discuss it as a group for a while.
  3. We'll break up into small groups to work on solutions.  Sometimes we will have reports back to the group as a whole on what was found.
  4. Finishing and extending solutions will be carried out as the homework assignment for that week.

The two tools we will use are OpenCyc and Shaken, which run under Windows.  (Sorry, Mac fans.)  Details on how to download them are in the Resources page. 

Last edited 9/15/03, by KDF.  Back to CS395-KR home page.