Abstract: "Because knowledge acquisition is a very difficult process, some qualitative reasoning systems use deep knowledge representing principles. But using deep knowledge increases the complexity of reasoning because the grain size of reasoning that uses only deep knowledge is sometimes too...

Full description

Bibliographic Details

Format:	Book
Language:	English
Published:	Tokyo, Japan 1992
Series:	Shin-Sedai-Konpyūta-Gijutsu-Kaihatsu-Kikō <Tōkyō>: ICOT technical report 820
Subjects:	Knowledge acquisition (Expert systems)
Summary:	Abstract: "Because knowledge acquisition is a very difficult process, some qualitative reasoning systems use deep knowledge representing principles. But using deep knowledge increases the complexity of reasoning because the grain size of reasoning that uses only deep knowledge is sometimes too small. We therefore propose a method for generating knowledge that has a larger grain size. This method generates 'aggregated knowledge,' representing the behavior of large components, from deep knowledge representing the behavior of small components. The generation process consists of three analysis steps. The first is qualitative simulation to find all possible bahaviors of the target large component The second is to find all the possible states from these behaviors. And the last is to find the transitional order of those states. These steps generate aggregated knowledge that has existential conditions, relations, and transitional orders for each possible state. Such aggregated knowledge can represent all kinds of components and is useful in applying qualitative reasoning to large and complex systems.
Physical Description:	22 S.