Study On Improving Strategy Generating Capacity Of The Extension Strategy Generating System

With the development of the social economy development and the network of information technology, more and more information and knowledge appear, all kinds of system are more complex, more parameters to be considered, and there are more and more contradictions. How to use computer and network’s feature in various fields of strategy, has become the key point to improve the computer intelligence.Extenics is a cross subjects created by Chinese scholars, with its formal model to study things of expanding and transforming and the possibility of innovation and method, and to resolve contradictions.With the development of HowNet (Network) continues to improve, with HowNet’s knowledge resources as the foundation, through the knowledge of data description language (KDML) and other tools, it would be expected to let the computer automatically generates problem solving strategy. The study has groundbreaking scientific significance, and lay a solid foundation for future development problems of intelligent processing, expanding the application field of HowNet. it will have broad application prospects, and the distinct innovation.This thesis is supported by the National Natural Science Foundation (grant number:61273306)-"the study and the problems based on HowNet correlation computing system research" research.Firstly, this thesis introduces the extenics, the extension strategy generating system based on knowledge and HowNet theory.Then, this thesis introduces how to apply the genetic algorithm to generating strategy, avoid the combinatorial explosion problem resulted in an increase in computational complexity index.At the same time, how to use the word similarity computing method in HowNet to improve method for calculating the relationship of words and the base element correlation calculation method, and then establishing a computer method for target correlation and the correlation of the formation conditions, building calculation of correlation. Finally, through the VS platform, realizing extension strategy generating module and calculation module related problems, and the experimental results are analyzed.The innovation of this thesis lies in:One, the extension has many types of transform base, how to choose the transformation in the transformation base requirements, can not only ensure the effective strategy generation to solve contradiction problems enough, but also to avoid the combinatorial explosion problem resulted in an increase in computational complexity index. For this, the genetic algorithm and the extension improve the extension strategy generating robustness, scalability, generate strategies more effectively than the traditional extension strategy generating system.Two, combining with the related knowledge of HowNet, improve of calculation method for the relationship of words using word similarity computing method in HowNet, first build the base element correlation calculation method, and then establish a computer method for target correlation and the correlation of the formation conditions, at last,build calculation of correlation. Through the calculation problem of correlation, for two problem of A and B with positive correlation, can solve the problem of A and find the similar method to solve the problem of B, and for two problem of A and B with negative correlation, by solving the problem A, while with the reverse way to solve the problem of B. And the related problems can be classified, by induction, summarize the general characteristic, that such problems, find a general rule to solve these problems, and improve the strategy generating method, so that the contradiction between unresolved issues resolved, or make the problem solved to find a better strategy.