Generating Relations from XML Documents

This paper discusses several mechanisms for creating relations out of XML documents. A relation generator consists of two parts: (1) a tuple of path expressions and (2) an index indicating which path expressions may not be assigned the null value. Evaluating a relation generator involves finding tuples of nodes that satisfy the path expressions and are related to one another in a meaningful fashion. Different semantics for evaluation are given that take into account the possible presence of incomplete information. The complexity of generating relations from documents is analyzed and evaluation algorithms are described.