Clustering XML Documents by Structure

This work explores the application of clustering methods for grouping structurally similar XML documents. Modeling the XML documents as rooted ordered labeled trees, we apply clustering algorithms using distances that estimate the similarity between those trees in terms of the hierarchical relationships of their nodes. We suggest the usage of tree structural summaries to improve the performance of the distance calculation and at the same time to maintain or even improve its quality. Experimental results are provided using a prototype testbed.