Efficient Eager XPath Filtering over XML Streams

Abstract. We address the embedding existence problem (often referred to as the fil-tering problem) over streaming XML data for Conjunctive XPath (CXP). Ramanan (2009) considered Downward CXP, a fragment of CXP that involves downward navi-gational axes only, and presented a streaming algorithm which solves the problem in O(|P ||D|) time using only O(|P |height(D)) bits of space, where |P | and |D | are the sizes of a query P and an XML data D, respectively, and height(D) denotes the tree height of D. Unfortunately, the algorithm is lazy in the sense that it does not nec-essarily report the answer even after enough information has been gathered from the input XML stream. In this paper, we present an eager streaming algorithm that solves the problem with same time and space complexity. We also show the algorithm can be easily extended to Backward CXP a larger fragment of CXP.