Research On Data Warehouse System Key Technologies For Enterprise Group

Enterprise Group accumulated a lot of design, production, inventory, sale, purchase, finance and other business data in the operation process of information. How to convert these mass business data into decision-making information which has become a difficult and hot issue of Enterprise Group information, data warehouse system is considered to be the best solution.Enterprise Group data warehouse system is a complex system involving many complex concepts and techniques. In this paper, several key techniques have been studied for Enterprise Group data warehouse system. This paper research results provide a good reference for the implementation of Enterprise Group data warehouse system, and have important theoretical and practical significance. The research works of this paper are listed as follows:(1) Study on the concept and architecture of data warehouse system for Enterprise Group. We give the definition of data warehouse system for Enterprise Group. We put forward the concept of unified views model, and propose the data warehouse system architecture based on the unified views model.(2) Study on data warehouse system ETL technology. We give a new data warehouse ETL architecture based on the unified views model, present an ETL process modeling and constructing approach based on the unified views model. At the same time for data warehouse ETL tasks scheduling problem, we take data warehouse ETL tasks minimum execution time as scheduling objective, establish ETL tasks scheduling model, propose a same layer division genetic algorithm for solving the scheduling model.(3) Study on Enterprise Group data warehouse technology. We provide a distributed data warehouse hierarchical structure for Enterprise Group, propose a distributed data warehouse model for Enterprise Group, and summarize the implementation strategy of distributed data warehouse and key technologies. At the same time, the model-driven approach is applied to the data warehouse development.(4) Study on Enterprise Group OLAP technology. A model driven architecture software development method is applied to OLAP development. In data warehouse system unified modeling framework, we can tackle the design of OLAP at the conceptual level instead of the logic level, design OLAP PIM model at the conceptual level, and implement OLAP development through model transformations from PIM model to PSM model and PSM model to SQL Code.(5) Study on Enterprise Group data mining technology. We propose a decision tree classification improved algorithm based on sampling which can also be effective in large data sets and mine correct classification rule. We apply this decision tree classification algorithm to the key processes mining of production cost, mine out the key processes of the process route and classification rules affecting iron& steel enterprise cost. At the same time, for association rule mining in large data sets, we put forward the trifurcate linked list storage structure of directed itemsets graph, and provide the association rule mining improved algorithm based on directed itemsets graph. Through customer data association rules mining applications for DongBei special iron& steel Enterprise Group, we can mine out customer's buying behavior and potential demand.(6) Study on Enterprise Group decision support technology. We provide Enterprise Group decision support system framework based on distributed data warehouse. Through defining decision schema layer and the decision task model layer, adopting hierarchical strategy, we can reduce the complexity of decision support system. Using object-oriented component development approach, we can ensure the effective integration of data and decision algorithm which enhance system's reusability and scalability.