Assumptions and limitations of a problem-solving method: a case study

The paper attempts a step in the direction of competence theories of reusable problem-solving methods (psm's). In fact, we examine a variant of the psm propose-and-revise. The variant was developed as a solution to the current Sisyphus task (VT-task). We decompose the entire method into its individual subtasks and examine each subtask for its underlying assumptions concerning the available domain knowledge. By doing this, we determine assumptions and limitations of the entire method. In addition, we examine how different control flows between these subtasks can influence the efficiency and effectiveness of the method. Again, we show how these differences are related to the assumptions of the method. Making these assumptions explicit defines interesting goals for validation and verification efforts. First, we show that under some circumstances two psm's behave equally well or that one is more suitable than another. Second, we have to prove for a given domain knowledge whether it fulfils the assumptions of a selected problem-solving method. We believe that such studies are necessary to allow the reuse of psm's and, conversely, that the reuse of psm's justifies the effort of such activities. (orig.)Available from TIB Hannover: RN 3492(312) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekSIGLEDEGerman