Abstract

This short position paper discusses the issues arising when the expectations of element retrieval are applied to heterogeneous document collections. One assumption of element retrieval strategies is that it is actually possible for searchers to specify the elements to be retrieved. As collections include an ever-increasing number of XML document types with varying schemas or DTDs, this knowledge cannot be expected on the part of searcher (unless one supposes the searcher to be omniscient), and in any case the complexity of queries must also grow monotonically with the number of types, making it increasingly difficult for the searcher to construct an element-oriented query.