Research And Implementation Of Several Aspects To Improve ESGS

With the scientific progress, intelligent tools are playing a more and more important role in national economy and social life. While the key factors for deciding how intelligent of a machine is its capability to resolve contradictory problems. Extension Strategy Generating System (or ESGS, for short) is a computer system, which based on the theory and method of extenics, aimed at generating strategy for resolving contradictory problems. Study on ESGS is very meaningful for the development of decision science and artificial intelligence. Current development technology of ESGS is not yet mature, therefore this paper studied several aspects to improve it.To effectively solve contradictory problems, we have to build a right extension model. Current ESGS generally uses the intelligent agent technology, guides the user to input parameters of the problem to establish the extension model. When there are more parameters, the design of input interface is difficult to reuse. When the text input by the user is slightly longer, the system cannot understand user requirement well, which leads to low efficiency of modeling. Therefore, this paper puts forward an approach by using information extraction technology, to build the extension model automatically from user requirement sentence.Basic-element is the logic cell to describe information and knowledge in extenics. Current knowledge in ESGS are expressed by basic-element or by the relations of the basic-element based on extension rules. When building knowledge base with relational database, knowledge is stored in a fixed schema, makes it difficult to extend. Besides, complex knowledge expressed by compound elements is not yet studied. Therefore, this paper studied representation and preservation of extension rules knowledge with non-relational database MongoDB.To solve contradictory problems, current ESGS generally preserve the extension transformations in database or hard-coding into the applications. This approach cannot make use of knowledge base when analyzing user’s problem, which restricts the extension transformation’s type and quantity, makes it hard to generate effective strategy. To improve the situation, this paper puts forward a correlation analysis algorithm which can take advantage of the correlation knowledge.The innovation of this paper:(1)By using the information extraction technology, successfully transformed the user requirement sentence into extension model. This approach makes ESGS to get user problem more efficiently.(2)Put forward the method of representing and preserving extension rules knowledge in non-relational database MongoDB, which improves the expansibility and operability of knowledge base.(3)Put forward the algorithm to make correlation analysis of incompatible problems. The algorithm can build correlation tree according to the correlation knowledge, execute and calculate the result of extension transformations, which makes correlation analysis more flexible.This is the research results of National Natural Science Fund Project "strategy generation method based on extenics and HowNet and system research" (No.: 61273306).