Research Of Improvement On Certain Theory

Expert systems have been popular for about thirty years and applied successfully in manydomains. Nevertheless, problematic expert system design issues remain-uncertaintymanagement is case in point. The knowledge representation and its reasoning is one of themost important parts since Expert System was born. There are two international conferencesabout theory and application of uncertainty reasoning every year. And many publications,papers, magazine and monographs are published. Some excellent experts such as Shotliffer,Duda, Shafer, Heckerman, Pearl, Dubois, Prade, Yager, etc work hard in this area and gotsome effective conclusions. At home, researchers of Jilin University was the earliest one toengage the expert systems and gained some harvest. Other organizations, such as TsinghuaUniversity, HeFei University of Technology and so on, are working hard in this area.Uncertain Reasoning is a very important research area of artificial intelligence. So far, theresearch about uncertain reasoning is mainly used to build reasoning model which is basedprobability theory, fuzzy mathematics or rough theory. Certain theory is the best convenientand practical among all of them.In part one ,this paper introduces the relationship between the uncertain reasoning andartificial intelligence,expert system and knowledge engine. In part two, it shows theactuality,research of purpose and the significance of the uncertain theory. There are somecontents on the basic mathematics (probability theory,fuzzy mathematics. etc . ) and someuncertain reasoning models such as Bayes network,the subjective Bayes theory,certaintheory,D-S theory and possibility theory in part three. And we analyses these theories basedupon many reasoning models and summarize the approach. In part four, it introduces a newknowledge representation and its reasoning algorithm is represent in part 5. The experimentdata is given to be used to analysis and compare three-way theories. From part 6, we can geta better result based on the new theory which is brought forward in part 5. Based on certaintheory, evidence representation and knowledge representation about uncertain reasoning areboth improved. And in this paper, a new reasoning arithmetic is given based on the aboverepresentation of evidence and knowledge. It is used to resolved some problems whichexited in old reasoning arithmetic.