| Abstract
|
Controlling inference in Artificial Intell … Controlling inference in Artificial Intelligence (AI) systems that usedeclarative representations has always been a key issue in AI research. Theproblem arises because the problem-solver's search is unfocused and considersmany irrelevant facts. In the past, meta-level reasoning and creatingabstraction hierarchies have tried to deal with this problem. The notion ofirrelevance plays an important role in both fields. A significant type ofmeta-level control knowledge are statements about irrelevance of entities inthe representation to problem-solving goals. Given a goal, it is possible tostate that some entities in our KB are irrelevant to it and should thereforenot be used in the search for its solution. In the case of abstractions,defining the abstraction hierarchy is equivalent to stating what aspects ofthe domain are relevant at each level. I describe a framework in which meta-level control facts about (ir)relevance can be stated and subsequently used bya problem-solver. These control facts, called relevance-claims are stated ina declarative fashion, much the same way as the knowledge about the domain.The claims state relevance between problem-solving goals and entities in theKB. Their meaning is defined by the way they constrain the possible space ofdeductions that the problem-solver considers. The framework will be studiedand implemented on a device-modeling system and on the Cyc system. ice-modeling system and on the Cyc system.
|
