Mining Historical Xml

Nowadays the Web poses itself as the largest data repository ever available in the history of humankind (Reis et al., 2004). However, the availability of huge amount of Web data does not imply that users can get whatever they want more easily. On the contrary, the massive amount of data on the Web has overwhelmed their abilities to find the desired information. It has been claimed that 99% of the data reachable on the Web is useless to 99% of the users (Han & Kamber, 2000, pp. 436). That is, an individual may be interested in only a tiny fragment of the Web data. However, the huge and diverse properties of Web data do imply that Web data provides a rich and unprecedented data mining source. Web mining was introduced to discover hidden knowledge from Web data and services automatically (Etzioni, 1996). According to the type of Web data, Web mining can be classified into three categories: Web content mining, Web structure mining, and Web usage mining (Madria et al., 1999). Web content mining is to extract patterns from online information such as HTML files, e-mails, or images (Dumais & Chen, 2000; Ester et al., 2002). Web structure mining is to analysis the link structures of Web data, which can be inter-links among different Web documents (Kleinberg 1998) or intralinks within individual Web document (Arasu & Hector, 2003; Lerman et al., 2004). Web usage mining is defined as to discover interesting usage patterns from the secondary data derived from the interaction of users while surfing the Web (Srivastava et al., 2000; Cooley, 2003). Recently, XML is widely used as a standard for data exchanging in the Internet. Existing work on XML data mining includes frequent substructure mining (Inokuchi et al., 2000; Kuramochi & Karypis, 2001; Zaki, 2002, Yan & Han, 2003; Huan et al., 2003), classification (Zaki & Aggarwal, 2003; Huan et al., 2004), and association rule mining (Braga et al., 2002). As data in different domains can be represented as XML documents, XML data mining can be useful in many applications such as bioinformatics, chemistry, network analysis (Deshpande et al., 2003; Huan et al., 2004) and etcetera. BACKGROUND