Research And Application Of Knowledge Warehouse System Of CRM Based On Knowledge

With the application of the CRM system in the business enterprise, more and more customer data was piled up in the business database of business enterprise, because of the limit of technique and idea, a lot of information and customer knowledge in customer data can not be dug out quickly and availably, how to convert these data to the information that is useful to business enterprise and make it could be shared with business enterprise have already become a bottleneck problem of CRM system. Therefore, this paper put forward the thought of applying knowledge management to the customer relationship management to carry on a research, apply the technique of knowledge discovery in the knowledge management field to analyze and dig out the latent and valuable information and knowledge, which will be used to support decision, improve customer's satisfaction, strengthen customer's reservation, enhance the profitability of business enterprise .The paper systematically studied the related theory and method of customer relation ship management based on knowledge management, did an initial study on the carrying out of system technique of KM–CRM, had important theoretic and practical meaning .To Begin with, the background of thesis researched, conceptions, characteristics, internal and overseas status quo about the customer relationship management, the knowledge management and the game theory are summarized, the deficiency in this area is discussed and the thought of customer relationship management based on knowledge management is put forward. At last, theoretical and practical significance of the thesis is analyzed All that mentioned above form the base of the research in this dissertation.A systematically study on the knowledge warehouse of CRM system is done and the knowledge warehouse system is put forward based on the ontology. The paper sets up a knowledge model based on the ontology, and introduces the concretely process of setting up the model by taking the semantic model of customer ontology as an example .The paper defines the process of ontology mapped. The similarity is calculated in four respects, such as syntax, attribute, structure and instance, after that, the process that maps the source ontology to the object ontology is realized. In the process to deduce the case knowledge in knowledge warehouse, the paper adopt the method of modeling based on the ontology, the case ontology can be a kind of general express model of case knowledge, not only can carry on an effective expression to the case knowledge, but also can express the characteristic of case, support the search based on the contents.The content and characteristics of customer segmentation are analyzed in detail,this dissertation broadens the quota system in traditional business environment, sets up the dynamic synthesis quota system in terms of customer value ,and uses the layer analysis method to make sure the weighs of index. The paper aims at the model of knowledge discovery of customer segmentation, uses the single classification machine, and the single classification methods has some limit, so the combine classification methods according to of the SOM& SVM is put forward. By comparing the result of experiment, we can find that the combine model has a good function to improve the efficiency of classification.In order to support the analysis of cross sale in customer's behavior, to aim the disadvantage while the existing classic association rules mining method producing rules, an association rules mining method is proposed based on frequent pattern tree with profits constrains. The method compresses the important information in originality trade database into a data structure named frequent pattern tree (FP-tree) to reduce the searching space of data. Its characteristics is that not a great deal of candidate item sets was created in the process of mining, times of scanning database were effectively reduced, the efficiency of mining rules was greatly increased, and a great quantities useless rule was avoided producing by the setup of profits constrains. At last the validity and practicability was verified by an instance.To aim the disadvantage of BP neural network training algorithms, we put forward an improved measure in which genetic algorithm is used to optimize neural network, the characteristic of genetic algorithm is used to seek excellences through overall situation to make up the disadvantage of BP algorithms which slows convergence speed and makes the result be easy to fall in to local least spot. So we can achieve the intention to seek excellences quickly and forecast accurately. We also improve the traditional genetic algorithm before optimizing neural network, the improved genetic algorithm has better convergence then unimproved one and can resolve the problem of earliness and convergence better. At last the feasibility and validity was verified by an instance.In order to provide computer-aided support for the custom relationship management, the KCRM system was designed and developed according to knowledge management. The architecture and function modules of the system were presented. Finally, an example was given to show the realization process of the system.