面向數(shù)據(jù)資源的認(rèn)知圖推理機(jī)制研究與實(shí)現(xiàn)

1、重慶理工大學(xué)碩士學(xué)位論文面向數(shù)據(jù)資源的認(rèn)知圖推理機(jī)制研究與實(shí)現(xiàn)姓名：田華庚申請(qǐng)學(xué)位級(jí)別：碩士專(zhuān)業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：陳莊2011-06-01ABSTRACT IIABSTRACT Cognitive map is a tool of soft computing and a novel method for knowledge management and knowledge representation, and it also

2、 can be used to represent the causality between entities in related fields. Cognitive map inference is the process of reasoning from known knowledge to unknown knowledge based on cognitive map. Cognitive map can be used

3、to represent knowledge by map directly，to model the complex system, and it has a powerful ability of numerical reasoning. Cognitive map reasoning can be used to predict the result of specific behaviors or to find the rea

4、sons of the facts, thus, it is considered as a hot research field in artificial intelligence these years. However, the previous research depended heavily on the expertise and ignored the objective data resource, which m

5、ay lead to the phenomenon of information loss. While the current research focused on the representation of cognitive map, the foundation of cognitive map, and the application of cognitive map, etc. Besides, the research

6、on cognitive map inference was so few, and lack of comprehensive and systemic research about inference mechanism and inference strategy. Firstly, we deeply analysed the development of cognitive map theory and the applica

7、tion of cognitive map, pointing out the problems of the contemporary study. Then we described the current methods of cognitive map inference (matrix reasoning, hierarchical reasoning, and rule-based reasoning included) e

8、laborately and analyzed the shortcoming of each method with living examples. Furthermore, we mainly studied the matrix reasoning. In this paper did intensive study on transformation function of cognitive map reasoning, a

9、nd analyzed the reasoning ability of binary transformation function, three-valued function and the sigmoid transformation function comparatively. The characteristics and the application scope of these three transformatio

10、n functions were summarized. Based on the previous study, we analyzed four key issues in the process of reasoning (how to get the initial conditions for reasoning, how to set the reasoning model, the choice of transforma

11、tion function, the judgement of the final state), and proposed a cognitive map inference algorithm. Combined with data preprocessing techniques, we used Visual C++ to implement “cognitive map inference system based on da