MINING TREE-BASED ASSOCIATION RULES FROM XML DOCUMENTS

The increasing amount of XML datasets available to casual users increases the necessity of investigating techniques to extract knowledge from these data. Data mining is widely applied in the database research area in order to extract frequent correlations of values from both structured and semi-structured datasets. In this work we describe an approach to mine Tree-based association rules from XML documents. Such rules provide information on both the structure and the content of XML documents; moreover, they can be stored in XML format to be queried later on. The mined knowledge is approximate, intentional knowledge used to provide: (i) Quick, approximate answers to queries and (ii) Information about structural regularities that can be used as data guides for document querying. A prototype of the proposed system is also briefly described.