A New Ranking Scheme and Result Representation for XML Information Retrieval Based on Benefit and Reading Effort

Elements of XML documents greatly vary in size and may nest each other. XML information retrieval (XML-IR) systems are required to take these nature of XML documents into consideration. Because of these natures, top-k search is not suitable for XML-IR. We introduce new concepts of benefit and reading effort of elements. By using these concepts, we propose a new dynamic ranking method that enables us to browse search results of XML-IR systems efficiently. We also study how to display the search results of XML-IR systems. If XML document was original composed of paper image pages, such as scholarly articles or books, it would be natural to display result elements by overlaying them on the physical layout of pages in the user interfaces. We propose methods for displaying the result of XML search of scholarly articles and ranking methods based on page units.