Building and searching an XML-based corporate memory

Abstract

No matter who uses a corporate memory or how it is constructed, information search through that memory should be efficient and effective. In particular, it should adapt to the users' needs, activities, and work environments. For a document-based corporate memory distributed through the Web, which is our research area, these requirements raise two main questions: how will we describe the documents that will be part of this corporate memory? How can we improve their retrieval on the Web, particularly by taking advantage of knowledge models? To answer these questions, we have developed a strategy that applies the XML (Extensible Markup Language) metalanguage to knowledge management for an enterprise. In this case, an enterprise is an actual company or a virtual enterprise made of distributed organizations and teams of people who work together online. We have implemented this strategy in the Osirix (ontology-guided search for information retrieval in XML documents) system. Because Osirix performs search based on the semantics of the documents, it can find more relevant documents than classic search engines can.