Embedding Similarity Joins into Native XML Databases

Similarity joins in databases can be used for several important tasks such as data cleaning and instance-based data integration. In this paper, we ex- plore ways how to support such tasks in a native XML database environment. The main goals of our work are: a) to prove the feasibility of performing tree similarity joins in a general-purpose XML database management system; b) to support string- and tree-based similarity techniques in a unified framework; c) to avoid relying on special data preparation or data structures to support simi- larity evaluation, such as partitioning or tailor-made index structures; d) to achieve a seamless integration of similarity operators and the existing database architecture.