Integrating Content and Structure into a Comprehensive Framework for XML Document Similarity Represented in 3D Space

Abstract

XML is attractive for data exchange between different platforms, and the number of XML documents is rapidly increasing. This raised the need for techniques capable of investigating the similarity between XML documents to help in classifying them for better organized utilization. In fact, the idea of similarity between documents is not new. However, XML documents are more rich and informative than classical documents in the sense that they encapsulate both structure and content; on the other hand, classical documents are characterized only by the content. According, using both the content and structure of XML documents to assign a similarity metric is relatively new. Of the recent research and algorithms proposed in the literature, the majority assign a similarity metric between 0.0 and 1.0 when comparing two XML documents. The similarity measures between multiple XML documents may be arranged in a matrix whereby data mining may be done to cluster closely related documents. In this chapter the authors have presented a novel way to represent XML document similarity in 3D space. Their approach benefits from the characteristics of the XML documents to produce a measure to be used in clustering and classification techniques, information retrieval and searching methods for the case of XML documents. We mainly derive a three dimensional vector per document by considering two dimensions as the document’s structural and content, while the third dimension is a combination of both structure and content characteristics of the document. The outcome from our research allows users to intuitively visualize document similarity.