Fusion Approaches for Mappings between Heterogeneous Ontologies Thomas Mandl, Christa Womser-Hacker University of Hildesheim, Information Science, Marienburger Platz 22 D-31141 Hildesheim, Germany {mandl, womser}@rz.uni-hildesheim.de Abstract. Ordering principles of digital libraries expressed in ontologies may be highly heterogeneous even within a domain and especially over different cultures. Automatic methods for mappings between different ontologies are necessary to ensure successful retrieval of information stored in virtual digital libraries. Text categorization has discussed learning methods to map between full text terms and thesaurus descriptors. This article reports some experiments for the mapping between different ontologies and shows further that fusion methods which have been successfully applied to ad-hoc information retrieval can also be employed for text categorization. 1 Introduction Ontologies are organized collections of concepts. Their structure expresses a certain view on the world or the domain. A considerable number of ontologies exists in many domains. This especially applies for different countries and cultures. Many ordering systems have concepts in common, however, because they arose within a certain context in response to special demands they may organize them differently. Although this situation is quite natural and not likely to change, it complicates or even prevents the successful communication between communities using different ontologies. The same is true for the exchange of documents between different groups within one digital library. The worldwide exchange of documents involves ontologies from different cultures which are sometimes organized extremely different. Due to the ever growing amount of knowledge available electronically, automated solutions need to be found for these mapping problems. Text categorization between ontologies or terminologies seems to be the most appropriate technology for this task. Mostly, text categorization assign documents to predefined categories based on a full text analysis [27]. The text is indexed with standard information retrieval methods and represented by weights assigned to words or terms based on their frequency of occurrence. These terms can be regarded as features. In the same manner, terms from a controlled vocabulary like an ontology can serve as features for a text. Thus, the task for text categorization based on full text terms is equivalent to text categorization based on descriptors from ontologies. There are also applications for the second case [4]. In: Constantopoulos, Panos; Sølvberg, Ingeborg (eds.): Research and Advanced Technology for Digital Libraries: 5th European Conference (ECDL 2001) Darmstadt 4.-8.9.2001. Berlin et al.: Springer [Lecture Notes in Computer Science 2163]. pp. 83-94.