Clustering of Web Documents with Structure of Webpages based on the HTML Document Object Model

Web mining is an emerging Data Mining arenathat usesvarious techniques to explore hidden patterns available within the WWW. Clustering has significant applications in Web mining, particularly in grouping Webpages based on their various properties. Literature suggests that clustering applied over Webpages is generally based on the contents of the availableWebpages, thereby focusing on text mining techniques only. But since unlike normal text documents Webpages are structured documents, there is a scope of exploring whether the structural properties of Webpages have any impact on their clustering. This paper aims to apply clustering on Web Documents based on DOM structure of Webpages, where the HTML-DOM structure of each Webpage has been represented as a string of characters, and then applying K-means clustering on the string representation. The same algorithm has been applied with four different distance measures on four different datasets. The clustering output in each case has been evaluated and the results have been compared.