A Decision Support System Based On Web Data Mining Research

Web data mining is utilizing the information processing method of data miningtechnology to discovery users’ interested useful information or knowledge from theweb information resources and web use records, the result can be used indecision-making. In the construction of business enterprise management informationsystem, the web data mining technology improves decision support processingcapability and discovers the potential market or customer base, so it greatly improvesthe quality of enterprise’s business and exactly predicts the future development trendof market and enhances the enterprise market competitiveness. The Web data mininghas the strong academic value and realistic significance. In view of the presentsituation, the mainly studies of this paper have following several aspects:At first, the paper systematically analyses several significant web data miningmethods such as cluster analysis, decision tree, neural network, support vectormachine and genetic algorithm. Combining with the characteristics of web datamining, the research emphatically analyses the principle of least square method andlinear regression method. The least square method is effectively combined withlinear regression method. With respect to the classification problem in data mining,linear regression prediction process is studiedSecondly, the least squares method and linear regression method is applied inthe system design based on the research of the algorithms. A prototype of decisionsupport system based on Web data mining is designed and implemented in Javalanguage. The function structure, the physical structure and other related aspects ofthe system are analyzed in the design and the process of data mining is presented inthis decision support system. The system has realized four major function moduleswhich are the system management, data management, data mining, auxiliary analysisso that it can satisfy the functional requirements based on web data mining.At last, the decision support prototype system was tested. The test target data isthe annual freight volume and the annual gross value of industrial output data of onefreight enterprise in recent years. Using the two types of data to forecast the futureinfrastructure investment, the experiment results show that the system has strongdecision-making support stability.