Fuzzy Approaches to Clustering XML Structures

Information on the hierarchical nature of XML data is essential in tasks of learning from XML document structures. Within this view, XML documents can be regarded as multi-represented data, which is the case when multiple representations of the document correspond to the generation of features at each structure level of the document separately. This chapter raises the importance of using fuzzy approaches to clustering XML document structures, since these approaches are shown to be effective in combining the information coming from different document representations that correspond to different hierarchy levels. For this purpose, we overview fuzzy encoding and similarity methods and present fuzzy clustering approaches which are particularly suited for being extended to handle XML document structures. We propose two different scenarios of fuzzy clustering of XML structures, which aim to either encode the document structure hierarchy using a fuzzy bag model or to specifically handle the multi-representation of the documents.