Research On Intelligent Integration Of Organizational Heterogeneous Knowledge For Decision Supporting

Managerial decision-making, as a knowledge intensive activity, is always challenging. To make high quality decision, decision-makers need to be good at integrating relative information and knowledge and be equipped with a great deal of experience. Especially under rapid change and complex competition envorinment, making decision correctly and timely is depand on having valuable information and knowledge. People often consider lack of information will lead to failure in make good decision. Actually, the real reason is that there is no knowledge which has integred view and is able to focaste future. Knowledge management has become a focused topic in managerial decision-making and decision support research area. There are great changes in support model of how managers utilize organizational data, inforamtion and knowledge during their decision making. Managers have to face to knowledge overload because of huge amout of knowledge, which leads to organizational knowledge's usage is very low. More seriously, most of organizational knowledge are segmented, isloated, heterogeneous, and distributed, and existed in different sources. This fact is also making providing effective dicision suppore more difficult. Obviously, effective knowledge integration is the basis and premise of solving those problems. Unfortunately, judgment of knowledge value, presentation and integration of tacit knowledge, as the key factor of knowledge integration, is still in progress.According to questions mentioned above, this paper has carried out some related research works:1. Building the theory framework of intelligent knowledge integration towards decision making. The research approach of this paper is different from traditional research path which focus on orgnizational behavior, is from naturalistic decision making perspective to discuss decision-maker centred knowledge integration. In order to discuss this question, this paper established a systematic theory framework which based on decision science, cognitive psychology, artificial intelligence, naturalistic decision making, and multi-agent system, and aimed to research intelligent knowledge acquisition, intelligent knowledge representation, intelligent knowledge evaluation, and intelligent knowledge apply. The theory framework is the foundation of this paper's theory analysis and model formulation.2. Propose a knowledge representation method to describe tacit knowledge depands on causuality relationship and turst relationship. By reviewing and analyzing knowledge representation methods in artificial intelligence, this paper proposed a tacit knowledge representation method based on fuzzy cognitive maps which is a kind of soft computing method. Futhermore, this paper introduce a novel concept-belief of knowledge, the concept is reflecting how expert is confident on they suggestion.3. Discuss the features of knowledge's value, giving the value funtion of knowledge towords decision making. The real knowledge value for decision-maker is whether the knowledge could decrease degree of uncertainty in decision situation. The function will calculate the concept value, causuality value and trust value to measure the contribution which the target knowledge may bring to decision problem. Based on the value of knowledge, decision-makers could filter knowledge and choose a right decision point. The result is lowering influence of knowledge overload.4. In order to overcome problem of access knowledge from heterogeneous knowledge sources, an intelligent approach of integrating knowledge based on multi-agent systems was introduced. Under the multi-agent system, this paper presented a new decision support scheme which implemented by sharing mind model in an intelligent agent in order to overcome bounded awareness. The agent uses the sharing mind model to integrate experience knowledge in the multi-agent system.5. The mechnism of form sharing mind model is according to the approach provided by Evidence Theory. The purpose is fusion experence knowledge from other agents (including intelligent agent and human).6. Carry out simulation experiments to evaluate and analysis the knowledge value model, multi-agent knowledge integration and shared mind model under Repast simulation system. Design a prototype of intelligent knowledge integration systems in multi-agent JACK platform.The new ideas contrubuted to theory and practice is listed below:1. Propose a method of decision problem formulazation according to key factors and their relationship of decision problem. The method converts decision prolem into decision situation.2. Provide causuality-trusty fuzzy cognitive map approach to representation tacit knowledge. The approach introduced influence of expert's confidence into knowledge properties, which would get better understand of tacit knowledge. The approach also make tacit knowledge could be analyzed quantitatively.3. Building knowledge value function based on Entropy Theory. The function includes three elements:concept value function, causualty value function and trust value function. Apply the function in processes of decision making would provide effective way to decrease knowledge overload.4. Establish a framework of intelligent knowledge integration based on RPD-BDI agent and multi-agent system techniques and proposal a novel RPD-CBR reasoning method of agent. This paper introduced individual mind model and sharing mind model to construct agent's knowledge structure. Depand on sharing mind model, agents could absorb experience knowledge from other agents, which help a single agent overcome its bounded awareness.Above all, intelligent knowledge integration is a challenging and worthy of exploring. Few researchers have applied naturalistic decision making theory and multi-agent system technique togther to study this area. This paper has tried to discuss the theory foundation, key issues, and relavant technology. And hope to provide some benefits on integrating decision support systems and knowledge management.