Mining Tree-Based Frequent Patterns from XML

The increasing amount of very large XML datasets available to casual users is a challenging problem for our community, and calls for an appropriate support to efficiently gather knowledge from these data. Data mining, already widely applied to extract frequent correlations of values from both structured and semi-structured datasets, is the appropriate field for knowledge elicitation. In this work we describe an approach to extract Tree-based association rules from XML documents. Such rules provide approximate, intensional information on both the structure and the content of XML documents, and can be stored in XML format to be queried later on. A prototype system demonstrates the effectiveness of the approach.