Abstract: An approach for calculating the structural similarity between XML documents is proposed in this paper.The structural information of an XML document is captured with a structural summary tree (SST).By encoding elements as digital numbers, a SST is transformed to a digit-labeled tree.Digital numbers at different tree levels are concatenated to form a vector after the normalization process.Consequently, each XML document is represented as an m-dimension vector.The GA-based clustering algorithm is adopted since it is able to provide good results irrespective of the starting configuration.Experimental results show the effectiveness and scalability of the approach.

Key words: XML, GA, SST (structure summary tree), information retrieval, document clustering