This course will examine computational approaches to representing, reasoning with, and inferring scientific knowledge. We will begin by discussing the nature of scientific theories and how scientists discover them. The remainder of the course will cover approaches to automating key tasks that arise in every scientific field, including the induction of descriptive laws and the construction of explanatory models. Illustrative examples will range from the reconstruction of episodes from the history of physics and chemistry to the generation of new results in biology and Earth science. Each meeting, participants will read and discuss papers that describe computational responses to specific scientific problems. Our aim will be to understand the diverse set of computational methods that one can use to represent, utilize, and infer scientific knowledge in a variety of domains.
Course prerequisites include familiarity with concepts and techniques from artificial intelligence and basic knowledge of scientific method. Participants should be able to think computationally in terms of knowledge structures and the mechanisms that operate on them.
The primary emphasis of this course is on reading and discussion, with an occassional lecture. Students will complete a series of exercises that require creating scientific models and relating them to data. Grades will be based on these assignments (50 percent) and on contributions to class discussion (50 percent).
There does not exist a textbook on the content of this course, but we will read a number of chapters from Dzeroski and Todorovski's edited volume, Computational Discovery of Communicable Scientific Knowledge, which will appear shortly. Other readings will come from the broader literature on computational scientific discovery.
For more information, send electronic mail to email@example.com