| China's telecommunications industry faces fierce industry competitions. Although at present they have established a large number of computer application systems, such as billing, business, accounting, and network management, which significantly improve the production efficiency of telecommunication enterprises and provide an effective guarantee to meet high quality service. However, these systems still have many deficiencies, such as isolation system, data can not be good use, a lack of mechanism to dynamically capture market opportunities and detect existing problems, enterprises will not be able to make quickly correct decision in face of competition.At the same time, business model of telecom operators transform gradually from "technology-driven" to "market-driven" and "customer-driven". Mass marketing has lost its edge facing the diversified, hierarchical and personalized needs of customers. Marketing concept of insight marketing and precision marketing through in-depth data analysis based on customer information, value and behavior has gradually been accepted by major telecom operators. They also desire a lot to mining value from data of various application systems, which can reduce marketing costs and improve marketing effectiveness.The solution meeting needs of telecom is building data mining platform in telecommunications industry. There are still some limitations to use existing data mining technology in telecommunications industry directly, such as a long development cycle, not enough flexibility. SOA is the technical standards to build new generation enterprise applications. It can integrate existing systems and achieve flexible flow. In this paper, based on data mining requirements of telecommunications industry, combined with existing data mining technology, we put forward a SOA-based data mining of the telecom industry solutions which can meet integration, flexibility and customizability requirements of telecommunications data mining. The main research contents are as follows:1,Give a study on the basic principles and flow of data mining, do some research on the classification of data mining algorithms, and mainly on the algorithms of cluster analysis, correlation analysis, decision tree and neural network. Analyzing and summing up application areas of various data mining algorithms.2,Do some studies on the telecommunications data mining flow and architecture, analyzing the data mining needs of telecommunications and algorithm applications. Add business logic layer, model management layer and business data mapping module on the structure, which can implement a more flexible telecom data mining.3,Studies on basic principles, technology and implementation of SOA. Quoted and analyzed IBM SOA Foundation Architecture, and proposed several key issues of SOA technology.4,Dividing basic services of telecommunications data mining based on demand and architecture of telecommunications data mining. Based on the study of SCA, SDO, BPEL and other SOA technologies, put forward a solution of SOA-based data mining application platform in telecom.5,Under guidance of the solution, design, deploy and realize a user behaviour analysis case of telecommunications industry. Main achieves the application of user clustering analysis, and design solution of user clustering SOA services. At the same time, give a display of the application. |
