Based On The Conceptual Representation Of The Dynamic Semantic Networks And Infiltration Reasoning

Knowledge representation and knowledge-based reasoning are important parts in the field of artificial intelligence research. A reasonable knowledge representation method and reasoning mechanism can improve the problem solving capability of knowledge-based system, especially when the problem does not belong to a specific domain. However, most traditional knowledge representation methods and problem solving mechanisms can be applied successfully only in specific domain, and few of them can be applied to the domain-open problem solving. Knowledge-based problem solving requires a concept system which contains rich semantic, and this concept system, as well as its reasoning mechanism is also the core of human's cognitive system. Human's concept system, composed of concepts and their dependent relations, is considered as the source of human's cognitive ability. Therefore, we can improve knowledge-based system to solve domain-open problem by studying the concept system and reasoning mechanism of human. There are many excellent researches about human's problem solving mechanism in the field of neuroscience and psychology, and they are worth to be used for reference. For example, the neural network research and the semantic memory research. However, since the motive and method of these researches are different from artificial intelligence, so neuroscience and psychology do not take the storage of knowledge and the realization of algorithm into consideration. On the basis of these related outstanding research results, this paper describes the soaking reasoning mechanism of problem solving by means of dynamic semantic network (DSN). The paper presents the formalized definition of DSN, the regulation of constructing DSN, and the regulation of soaking reasoning which based on DSN. Meanwhile, it also presents other related formalized algorithm. Finally, the paper describes detailed motivation-spread algorithm which based on DSN. In addition, the paper takes a practical problem solving as example to realize the motivation-spread algorithm in computer, so it simulates soaking reasoning process of the problem solving. This research contributes to the improvement of reasoning and problem solving ability in knowledge-based system.