Abstract:
ID/IDREF is an important and widely used feature in XML documents for eliminating data redundancy. Most existing algorithms consider an XML document with ID references as a graph and perform graph matching for queries involving ID references. Graph matching naturally brings higher complexity compared with original tree matching algorithms that process XML queries. In this paper, we make use of semantics of ID/IDREF to reduce graph matching to tree matching to process queries involving ID references. Using our approach, an XML document with ID/IDREF is not treated as a graph, and a general query with ID references will be decomposed and processed using tree pattern matching techniques, which are more efficient than graph matching. Furthermore, our approach is able to handle complex ID references, such as cyclic references and sequential references, which cannot be handled efficiently by existing approaches. The experimental results show that our approach is 20-50% faster than MonetDB, an XQuery engine, and at least 100 times faster than TwigStackD, an existing graph matching algorithm.
