Customer Knowledge Mining Based On Semantic Integration

Customer knowledge has become an important competition resource of enterprise under e-commerce environment. How to better understand customer preferences, stimulate and meet customer demands, is the key problem of enterprise to win in the incandescent competition environment. Knowledge mining as an important concept for finding potentially valuable knowledge from large amounts of data, gets high attention of enterprises and researchers. The customer knowledge mining based on semantic integration is studyed, because the semantic heterogeneity in customer data has become the biggest obstacle to reduce the efficiency of knowledge mining tool and the effectiveness of customer knowledge model. The dissertation is composed of Introduction and other five chapters. The main contents are as follows:In the Introduction, the background and significance of this dissertation are explained. The related research status about semantic integration and customer knowledge mining at home and abroad are reviewed. In particular, these research results about the semantic integration, the comprehension of customer knowledge, the construction of knowledge management framework based on ontology, the plight and solution of knowledge mining, give a certain theory of inspiration and guidance for finding the entry point of this dissertation. The main research ways and contents of the dissertation are presented.In chapter one, the basic theories and technologies of this dissertation are overviewed. The main contents of enterprise information integration is described, which include the connotation of heterogeneous data and the goal of information integration, the traditional method of enterprise information integration, the semantic integration of enterprise information based on ontology. The main contents of customer analysis theories about the life cycle of customer relationship, the customer value, the customer loyalty and the customer satisfaction are described, which lay the basis of theoretical analysis about the formulation of customer knowledge mining tasks. The main contents of supporting technologies are reviewed, which include knowledge mining methods and technologies, data warehouse technology and ontology technology.In chapter two, the stratege and method of semantic integration about customer data based on enterprise ontology are proposed. The new idea of enterprise global ontology construction is put forward from software engineering perspective, the ontology construction is divided into planning, analysis, design, implementation and operation of the five stages using structured approach, and the ontology is modified in the analysis stage using prototyping method. The semantic integration strategy and method of relational database and web text based on enterprise global ontology is proposed. The process of relational database semantic integration using hybrid ontology method includes ER schema elements identification based on reverse engineering, local ontology learning based on ER schema elements, metadata generation based on ontology mapping. The process of web text semantic integration using single ontology method includes discovering concepts and conceptual relations in web text using clustering and association rule mining technologies.In chapter three, the customer knowledge mining model based on semantic integration is proposed. This model includes six circulating modules, they are valuable customer data collection and semantic integration, semantic integration results analysis, customer knowledge mining process based on semantic integration, customer knowledge pattern processing, customer knowledge pattern storage and application. Among them, we elaborate with emphasis the contents of semantic integration of customer data collection, customer knowledge mining task defining, and customer knowledge mining process based on semantic integration.In chapter four, an experiment to test the former research results is conducted. In this experiment we collect some valuable customer web log data in the field mobile phone, and get web texts and interactive feature sets. Then the semantic analysis and annotation of web text is implemented based on phone products ontology, and concepts and conceptual relations in web texts are given a certain weight according by interactive feature sets. On this basis, we get customer interest knowledge patterns with hidden semantic relations using the Apriori algorithm, and test the validity of these patterns in the basis of web text test sets.In chapter five, the main contents and conclusions of the dissertation are concluded. The disadvantages are reflected. The future research is projected.The dissertation is sponsored by the major project of Humanities and Social Science Key Research Center of Ministry of Education with the granted number08JJD870225.