In the paper we present a methodology for the semiautomated extraction of ontological knowledge from XML data sources in a given domain. We consider an interconnection scenario over the Web, where various independent organizations use XML as a data interchange language for conducting business activities. The resulting ontology is organized into concepts and concept relationships at different levels of detail. Semantic links within and between the different levels are studied in order to capture and formalize semantic relationships between concepts to support multiple views on the ontological knowledge.