Frequent Mining on XML Documents

Introduction

With the emergence of XML (eXtensible Markup Language) standardization, XML documents have been widely used and accepted in almost all the major industries. This has resulted in a large amount of XML documents. In order to discover useful knowledge from the huge collections, various data mining techniques have been used (Nayak, 2005). One of the popular data mining techniques is frequent pattern mining on XML documents (Tan, Dillon, Hadzic, Chang, & Feng, 2005), which has created a great deal of interest among researchers due to its potential of application in diversified fields. Some of them include data warehouse integration (Termier, Rousset, & Sebag, 2002), bio-informatics (Hadzic, Dillon, Sidhu, Chang, & Tan, 2006b; Zaki, 2005), network analysis and information retrieval . Furthermore, XML frequent pattern mining serves as an important function for other data mining tasks such as classification (Zaki, 2005), clustering (Kutty, Tran, Nayak, & Li, 2007; Nayak & Iryadi, 2007) and association rules mining(Rusu, Rahayu, & Taniar, 2007).

In spite of its potential, there are major challenges that are yet to be fully addressed while mining XML documents in comparison to the established field of frequent pattern mining from relational databases. Firstly, unlike the relational databases which has structured or the unstructured data such as text, sound or image files, XML documents model the hierarchical relationships between data items. XML data represent semi-structured format consisting of values and their relationships. An XML document contains tags and the data enclosed within those tags. The tag, that describes the element name, includes the semantics in the form of text data. The tags define the structure of an XML document showing the relationships between elements of the document. Hence, while mining the XML documents, the structure of the data items has to be preserved and taken into consideration. This contributes to the second challenge that is, dealing with huge volume of data compared to structured data¹ due to the storage of the structural information (Nayak, 2005).

This chapter presents a discussion of the current trends in frequent pattern mining on XML documents, addresses the related issues and challenges and finally presents the applications and opportunities for this type of mining. The background of frequent mining is provided in Frequent Pattern Mining section. The XML Frequent Pattern Mining section covers the analysis of various types of frequent pattern mining. Also, we will present some of the potential opportunities of XML frequent pattern mining from a technological as well as an application-based perspective in Applications and Issues of XML Frequent Mining and highlight some of the issues and challenges.