Query translation from XPath to SQL in the presence of recursive DTDs

• Published in: The VLDB journal Vol. 18; no. 4; pp. 857 - 883
• Main Authors: Fan, Wenfei, Yu, Jeffrey Xu, Li, Jianzhong, Ding, Bolin, Qin, Lu
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer-Verlag 01.08.2009; Springer
• Subjects: Applied sciences; Computer Science; Computer science; control theory; systems; Database Management; Exact sciences and technology; Information systems. Data bases; Memory organisation. Data processing; Regular Paper; Software; XPath; Query translation; XML database; SQL; Recursive DTD; Capability index; Query formulation; Database query; Relational database; Rewriting; Very large databases; XML language; Document type definition; Xpath language; Optimization; Database
• ISSN: 1066-8888; 0949-877X
• DOI: 10.1007/s00778-008-0131-0

We study the problem of evaluating xp ath queries over xml data that is stored in an rdbms via schema-based shredding. The interaction between recursion (descendants-axis) in xp ath queries and recursion in dtd s makes it challenging to answer xp ath queries using rdbms . We present a new approach to translating xp ath queries into sql queries based on a notion of extended XP ath expressions and a simple least fixpoint ( lfp ) operator. Extended xp ath expressions are a mild extension of xp ath, and the lfp operator takes a single input relation and is already supported by most commercial rdbms . We show that extended xp ath expressions are capable of capturing both dtd recursion and xp ath queries in a uniform framework. Furthermore, they can be translated into an equivalent sequence of sql queries with the lfp operator. We present algorithms for rewriting xp ath queries over a (possibly recursive) dtd into extended xp ath expressions and for translating extended xp ath expressions to sql queries, as well as optimization techniques. The novelty of our approach consists in its capability to answer a large class of xp ath queries by means of only low-end rdbms features already available in most rdbms , as well as its flexibility to accommodate existing relational query optimization techniques. In addition, these translation algorithms provide a solution to query answering for certain (possibly recursive) xml views of xml data. Our experimental results verify the effectiveness of our techniques.