International Journal of Computational Linguistics Research Tadeusz Pankowski 1 1 2010 http://www.dline.info/jcl/fulltext/v1n1/5.pdf Discovering matches and mappings between heteroge neous, independently designed data sources, is a challenging issue in data exchange and data integration. To deal with the problem we enrich XML schemas with seman tic information from a domain ontology by annotating the schema. In this paper we discuss how the annotations can be used to establish matching and mapping formulas be tween XML schemas automatically and propose a set of rules to infer these formulas.