Qualitative Reasoning about Physical Systems: A Return to Roots

Seven years ago the Journal Artificial Intelligence published a special volume entitled "Qualitative Reasoning about Physical Systems" [3]---qualitative reasoning for short. This special volume proved to be a watershed for qualitative reasoning research, which inspired a decade of ferment and innovative exploration. Now qualitative reasoning has reached another watershed, by returning to its original roots of coping with fundamental engineering tasks. ~ As both editors feel that addressing real tasks'is crucial to the success of the qualitative reasoning enterprise, we decided that now would be a judicious time to recognize and explicitly encourage this focus--through this special volume. This introduction conveys our personal perspective on what qualitative reasoning is about, what are some of the common misconceptions, and where we would like to see the research headed. The motivation for qualitative reasoning arose predominantly from research on engineering problem solving, which sought techniques for automating engineering practice for a variety of important tasks--circuit analysis and parameter selection [15,42,41 ], the diagnosis and teleological description of bipolar amplifiers [11 ], tutoring systems for steam plant operation [ 19 ], the design of MOS memory buffers [48], and the interpretation of geological structures [38]. It quickly became clear that, if we want to capture the skills of an engineer or technician we must do far more than build bigger simulators (such as Spice [44]) or equation solvers (such as Macsyma [28])--the