User queries on extensible markup language (XML) documents are typically expressed as regular path expressions. A variety of indexing techniques for efficiently retrieving the results to such queries have been proposed in the recent literature. While these techniques are applicable to documents that are completely schema-less, in practice XML documents often adhere to a schema, such as a document type descriptor (DTD). In this paper, we propose schema-conscious path-hierarchy indexing of XML (SphinX), a new XML indexing scheme that utilizes the schema to significantly enhance the search process. SphinX implements a persistent index structure that seamlessly combines the schema information with standard B-tree technology, resulting in a simple and scalable solution. A performance evaluation over a variety of XML documents, including the Xmark benchmark, indicates significant benefits with regard to both index construction and index access.
展开▼
