非表示:
キーワード:
-
要旨:
Highly heterogeneous XML data collections that do not have a global schema, as
arising, for example, in federations of digital libraries or scientific data
repositories, cannot be effectively queried with XQuery or XPath alone, but
rather require a ranked retrieval approach. As known from ample work in the IR
field, relevance feedback provided by the user that drives automatic query
refinement or expansion can often lead to improved search result quality (e.g.,
precision or recall). In this paper we present a framework for feedback-driven
XML query refinement and address several building blocks including reweighting
of query conditions and ontology-based query expansion.We point out the issues
that arise specifically in the XML context and cannot be simply addressed by
straightforward use of traditional IR techniques, and we present our approaches
towards tackling them.
