Representation And Acquisition Of A Procedural Knowledge Interaction Based And Its Realization

Knowledge is so powerful that diverse forms of knowledge result in various intelligent systems. Therefore, the representation of knowledge plays an important role in the acquirement, management and implementation of knowledge. The major objective of this thesis is to establish a system model based on procedural knowledge. Using the archived system model, the form of procedural knowledge can be represented with that of synthetic language. Moreover, as an application and validity of the system model, an expert system for work-order-sheets of electric power has been developed. Briefly, the prime contributions of this thesis can be generalized as the following: 1 . A functional language system with embedded construction for representing procedural knowledge is designed and implemented. This system consists of two parts: a language editor based interactive syntax-directed in web and a language interpreter. The working efficiency of the system depends strongly on the optimal design of the intermediary language.2. A storage mapping of G-tree, a type of syntax trees, for relational databases is proposed. As a form of context-free language, the storage mapping has several advantages, such as identical representation, simple sentence construction, and quick interpretation. The intermediate representation of sentence as context-free language in the storage mapping is sent to abstract-machine, and it is straightforwardly interpreted by the abstract-machine without processing. A language editor for sentence of context-free language in the storage mapping based on interactive syntax directed and web is designed and implemented. The representation and storage mapping is used in functional language system for describing procedural knowledge.3. Because acquire-knowledge is a bottleneck, inductive acquisition of knowledge also is addressed. Using the method of machine learning, the synthetic language is induced from a specific language of nature-language. The learning of synthetic-language-program is realized. An algorithm for reducing attributes in rough set theory is used to create sentence of function language. Because the domains of language in the system is defined on the relational database and specific-set, attribute reduction employing heuristic rule as domain knowledge can be used in cognizing procession.4. A kind of web models of software developing based on improved Petri Net is introduced. By means of the model, the extended finite automaton is converted into improved state charts, and then into multifunction views. The result is developed to a model of improved Petri Net, of which the description ability is much powerful and easy to meet the requirements of analysis and description. Applying this model, it only needs to establish separated components or sub-modules for further development of software and subsequent system refactoring.5 . An open and self-forming expert system of electric work-order-sheets based on B/S (Browser/Server) is presented. It can be utilized in transmission and distribution electrical maintenance and dispatching management system. By procedural describing knowledge and dynamic frame, theexpert system with strong adaptability is independent of electric-system connection and equipment. A complex inference engine for generating operation-item is also concerned. Additionally, a developing model for software workflow based finite automaton is discussed. This workflow management system can treat some sophisticated problems in equipment maintenance and management.