Study Of Artificial Intelligence-Based Knowledge Chain Model And Knowledge Acquisition And Presentation

"Study of Artificial Intelligence-based Knowledge Chain Model and Knowledge Acquisition and Presentation" is a branch of "Management Decision-Oriented Intelligent Systems Study of Isomeric Knowledge Representation and Knowledge Management" which is sponsored by National Science Foundation of China (NSFC, No: 70271002).Knowledge is more important than ever before in knowledge economy area, and is becoming a dominant element in supporting economy development. Today, there is a growing recognition in the business community about the importance of managing knowledge as a critical source for competitive advantage, and knowledge management has been widely adopted and applied in practices. Knowledge management is concerned with ensuring that the right knowledge is available in the right form to the right processors at the right time for the right cost. Execution of the KM activities undertaken in pursuit of this objective result in a panorama of knowledge chain(also called knowledge process). But at present, knowledge chain has not attracted enough attention in the KM field, and KM could benefit from studying of knowledge chain, which is the focus of the paper.As knowledge chain is a new research field, and current focus is mainly on conceptual model. This paper proposes a knowledge chain model based on artificial intelligence technology, which is the outcome of KM and AI. After discussing the proposed model in detail, this paper focuses on acquisition in knowledge chain.Knowledge acquisition is a bottleneck in the developing process of Knowledge-based System (KBS) or Expert System, and so does in knowledge chain. Knowledge chain could progress smoothly if and only if knowledge acquisition has been solved. The paper studies knowledge acquisition in two points of view, first the widely used technique called Structured Situation Analysis in expert system development is introduced;second, several automatic knowledge acquisition methods are proposed using AI technology, like artificial neural network (ANN) and decision tree (DT). As for ANN, the experiments conducted indicate that the Standard Backpropagation Network (SBP) does not work as well as revised BP which use Cross-entropy as error function. Also, this paper not only revises the RX algorithm in pruning and rule extracting process of knowledge acquisition, but also proposes two rule extraction method: Structured-based Rule Extraction from Neural Network (SRE for short), and Karnaugh Map-based Rule Extraction (KRE for short).It is difficult to represent knowledge in a uniform presentation, so the paper employs XML technology (XML Schema) to presentknowledge in order to simplify the store and search process in KC.In the end of the paper, several experiments are conducted using data from bank loan risk evaluation, which employ four different methods to acquire knowledge directly from data;these methods are RX, SRE, DT andKRE.In general, this paper has done a lot work both on theory and experiments, and due to the limitation of proper tools, the paper develops an Artificial Intelligence-based Knowledge Acquisition System (AIKA), and all the experiments are conducted under AIKA. Hence, it makes studying of knowledge acquisition easier than ever before.