Abstract
This paper presents two configuration problems: that of configuring computer hardware to meet a given specification, and an “engineer-and-made-to-order” problem in the domain of breathing air compressor production. We demonstrate how the different kinds of knowledge needed to solve problems in each domain (which we label factual, heuristic, and strategic) was captured for a proof
planning system. The systems perform favourably when compared with human experts. The ideas developed for the first domain (computer hardware) transferred remarkably well to the new domain (compressors), the time taken from knowledge acquisition to prototype being less than three months. We show that such systems are also easy to maintain, and to adapt to similar but distinct problems in the same domain. We end with suggestions of further domains and problems which we believe are amenable to the proof planning approach.
planning system. The systems perform favourably when compared with human experts. The ideas developed for the first domain (computer hardware) transferred remarkably well to the new domain (compressors), the time taken from knowledge acquisition to prototype being less than three months. We show that such systems are also easy to maintain, and to adapt to similar but distinct problems in the same domain. We end with suggestions of further domains and problems which we believe are amenable to the proof planning approach.
Original language | English |
---|---|
Publication status | Published - 1996 |
Event | The 9th Symposium and Exhibition on Industrial Applications of Prolog (INAP'96) - Hino, Tokyo, Japan Duration: 16 Oct 1996 → 18 Oct 1996 |
Conference
Conference | The 9th Symposium and Exhibition on Industrial Applications of Prolog (INAP'96) |
---|---|
Country/Territory | Japan |
City | Hino, Tokyo |
Period | 16/10/96 → 18/10/96 |