Research On Key Technologies Of Knowledge Excavation And Service Based On Heterogeneous Network

In order to meet the rapid development of modern service industry, especially to adapt to the development requirement of knowledge service scopes, the theories and technologies of heterogeneous network management, knowledge mining, knowledge management and knowledge service, they all need deep research. On the basis of heterogeneous network management, this paper deep studies the application theories and key technologies of knowledge mining and service. The research productions concern the innovation, development of knowledge service scopes and service supply mode.First, this paper summarizes the characteristics of a new generation of heterogeneous network and the development trend of network convergence; analyzes the main technical problems of heterogeneous network management and the main demand of building a large-scale heterogeneous network management system; evaluates the research actuality of key technologies for knowledge mining, knowledge management and knowledge service; indicates the shortcomings in these field, establishes the main contents and thinking of this article.In allusion to the main problems of knowledge management and knowledge service, especially the theory, model, method and mechanical design problems of knowledge discovery and knowledge expression, on the basis of a thorough study of natural language understanding, knowledge mining algorithm and knowledge resource model, we design a subject knowledge mining and resource management framework, and describe the function, flow and architecture design of it.In allusion to the difficult problems of integrated management for network having a large number of nodes and network equipment having various types in the heterogeneous distributed network management, on the basis of researching the Distributed Integrated Management Architecture (DIMA) and Self-Management Model, through analysis the problems of optical disjoint paths algorithm with QoS constrains, we design a service-oriented network comprehensive availability formula and related algorithm, which resolve some key problems of network independence management. In allusion to the pop research topic such as transforming the data advantages to the competition abilities of enterprises, on the basis of deeply analysis the evaluation method of text clustering analysis and research clustering validity functions, we propose a clustering algorithm based on genetic algorithm (GA) that can implement the customer behavior analysis. The features of customers are extracted according to the targets of customer subdivision, and secondly with the help of fuzzy dissimilarity matrix the relations between the huge amounts of modes of customer behavior are mapped into the two-dimensional matrix. With the presented clustering algorithm the customer behavior clustering analysis can be realized flexibly and quickly.By integrating the Web traffic agent filtering, AJAX, Comet, ICAP and other technologies, we propose a collaborative web page browse application based on Comet event mechanism. Human action is transported to a collaborative server by additional AJAX process, audience browser's reaction is done by Comet data receiving. By filtering the URL and JavaScript code in the original web page, the collaborative server can convert the ordinary single user oriented web page into multi-user collaborative web page without changing the media type of content.In the analysis of the Grid resource management framework based on Globus Architecture, comprehensively considering the aspects of basic infrastructure, information integration, virtual Ubiquitous Computing and information agent, we propose the knowledge management and service strategies of building a Grid-based ubiquitous knowledge service platform, to provide effective countermeasures for improving the knowledge resource management and enhancing the knowledge service capacity.