DTD-Based XML Indexing

Path expression is a common feature of XML query languages. Many indexing methods have been proposed. DTD can be of great help in XML indexing, but most techniques available now are generic to XML documents that are completely schema-less. Proposed in this paper is DBXI, a new method that takes advantage of information embedded in DTD for speeding up the process of XML path query. DBXI adopts a new coding scheme. With the help of DBXI, a path expression with N elements/attributes and one predicate restriction needs only zero or one structural join operations per XML document. For a path expression that does not match with any paths in XML documents, DBXI can give a judgment of no answer in much shorter time than those of indexing methods in existence. Experimental results demonstrate that DBXI can process path queries faster than Lore, SphinX and XISS do.