Item

Abstract

Information retrieval on XML combines retrieval on content data (element and attribute values) with retrieval on structural data (element and attribute names). Standard query languages for XML such as XPath or XQuery support Boolean retrieval: a query result is a (possibly restructured) subset of XML elements or entire documents that satisfy the search conditions of the query. Such search conditions consist of regular path expressions including wildcards for paths of arbitrary length and boolean content conditions. We developed a flexible XML search language called XXL for probabilistic ranked retrieval on XML data. XXL offers a special operator '$\sim$' for specifying semantic similarity search conditions on element names as well as element values. Ontological knowledge and appropriate index structures are necessary for semantic similarity search on XML data extracted from the Web, intranets or other document collections. The XXL Search Engine is a Java--based prototype implementation that support probabilistic ranked retrieval on a large corpus of XML data. This paper outlines the architecture of the XXL system and discusses its performance in the INEX benchmark.