Promoting the semantic capability of XML keys

Abstract

Keys for XML data trees can uniquely identify nodes based on the data values on some of their subnodes, either in the entire tree or relatively to some selected subtrees. Such keys have an impact on several XML applications. A challenge is to identify expressive classes of keys with good computational properties. In this paper, we propose such a new class of keys. In comparison to previous work, the new class of XML keys is defined using a more expressive navigational path language that allows the specification of single-label wildcards. This provides designers with an enhanced ability to capture properties of XML data that are significant for the application at hand. We establish a sound and complete set of inference rules that characterizes all keys that are implicit in the explicit specification of XML keys. Furthermore, we establish an efficient algorithm for deciding XML key implication.