Modelling stars using XML

We suppose collections of XML data described by Document Type Definitions (DTDs). This data has been generated by applications and plays a role of OLTP database(s). A star schema, a well-known technique used in data warehousing, can be applied. Then dimension information is supposed to be contained in XML data. We will use the notions of subDTD and view, and formulate referential integrity constraints in XML environment. We use simple pattern matching capabilities of current XML query languages for XML view specification and tree embedding algorithms for these purposes. A dimension hierarchy is defined as a set of logically connected collections of XML data. Facts may be also conceived as elements of an XML document. Due to the structural complexity of XML data the approach requires subtler formal model than it is done with conventional dimension and fact tables described by classical star schemes. In consequence, our approach captures more from heterogeneity of source databases than it is done in classical relational approaches to data warehousing.