The Reaserch And Design About Process Warehouse Used In Business Intelligence

The foundation of business operation are business processes, the level of process management will directly affect the efficiency of the process execution and the performance of enterprise. Business Process Management (BPM) and Business Intelligence (BI) are the information management methods commonly used in current enterprise. With the development of enterprises, the demand for exchange of information between different departments is increasing, it prompts companies to integrate different information systems, achieving information sharing within the enterprise. Meanwhile, because of the frequent cooperation between different enterprises, it is difficult to achieve communication of information without a standard. Faced with increasingly complex business processes, plus the amount of information increase constantly, it requires the combination of BPM and BI to improve the efficiency of enterprises, which is one focus of this article, called business process intelligence (BPI).In this paper, we integrated the existing process assessment system into the design procedure of BPIS, and data which worked from process mining or OLAP analysis is used for assessment system. According to their own requirement, users needs to choose the data after assessment, to further ensure the quality of mining and analysis results, making the whole system more complete. In order to cater to the process assessment, we rebuilt the initial process model, and defined the elements in, and used formal language described them. To better express the process effect of process warehouse for process log, we brought XES standard in storing process log data, and made a detailed research and analysis about matedata in process warehouse, and then analyzed the applications of XML Schema in metadata definition about process warehouse. On the basis of the existing theory, we proposed access-oriented process warehouse model(AOPWM) and gave the definition of its elements, explained the calculation method of metrics in fact table. We used multiple linear regression and gradient descent method to train the weight of process nodes, easy to compute the following customer satisfaction and the score in process assessment. Then we introduced two major analysis methods, OLAP and process mining. In the end, we improved a typical a algorithm in process mining field, solving the problem that can’t mining short cycle structure in a algorithm.End of the thesis, we applied AOPWM in the management of hospital process data. Through building a prototype system PAPWS realized process warehouse storing process log in Multidimensional Data Model. We described metadata in detail, and achieved computational algorithm of metrics in fact table, and made some OLAP analysis. Mining process log data using the improved a algorithm we abtained a It provided convenience for the following process assessment, making the result of data analysis support business decisions better.