ERIC Number: ED207583
Record Type: Non-Journal
Publication Date: 1976-Dec
Reference Count: N/A
Structured Planning and Debugging: A Linguistic Theory of Design. AI Memo 387.
Goldstein, Ira P.; Miller, Mark L.
A unified theory of planning and debugging is explored by designing a problem solving program called PATN. PATN uses an augmented transition network (ATN) to represent a broad range of planning techniques, including identification, decomposition, and reformulation. (The ATN is a simple yet powerful formalism which has been effectively utilized in computational linguistics.) PATN's plans may manifest "rational bugs," which result from heuristically justifiable but incorrect arc transitions in the planning ATN. This aspect of the theory is developed by designing a complementary debugging module called DAPR, which would diagnose and repair the errors in PATN's annotated plans. The investigation is incomplete as PATN has not yet been implemented, but sufficient detail is presented to provide a theoretical framework for reconceptualizing Sussman's HACKER research. Since a detailed study of planning and debugging techniques is a prerequisite for complete fulfillment of Dijkstra's objectives of program reliability, readability, portability, and so on, the theory is called "Structured Planning and Debugging" to emphasize its potential role in this enterprise. Fifty-seven references are listed. (Author/LLS)
Publication Type: Reports - Descriptive
Education Level: N/A
Sponsor: Advanced Research Projects Agency (DOD), Washington, DC.; National Science Foundation, Washington, DC.
Authoring Institution: Massachusetts Inst. of Tech., Cambridge. Artificial Intelligence Lab.