Research On A Semi-structured Data Model Based Frequent Patterns Mining

With the rapid development of Internet all over the world, finding regular contents and using available data become the hotspot, but the huge information and complexity of the information make information handling a little bit difficult. In order to solve the problem, producing semi-structured data model and Web Data Mining are the effective ways of the problem solving. Then, frequent pattern mining is a basic problem of the data mining area, and its research includes frequent pattern mining of various data. Its methods are widely used in other data mining arease. The topic of frequent pattern mining has attracted more and more researchers attention because of its basic nature and inner complexity.The concepts and present research status about data mining, Web data mining and XML are introduced in the thesis. The structure of semi-structured data and some semi-structured data models are analyzed. Semi-structured data model and xml data have some comparability, but xml is a data marked language, not a data model. Based on the requirement, a treelike object model named ATE is produced and it is used as data model when mining frequent patterns in xml.An algorithm named BATEMINER to mine frequent patterns in xml is produced. To some extent, ATE model reduces the data in xml, and uses less storage, so that decreases time in the mining process, thereby improves the efficiency of the algorithm.