Representing Research Activities in a Hierarchical Ontology

The present work is motivated by the problem of analyzingthe activities of an organization and representing them in ataxonomy of the domain as its hierarchical ontology. We focus onrepresenting the research activities of a Computer Science researchorganization in terms of the ACM-CCS taxonomy. We derive researchtopic clusters according to the similarity derived on the basisof profiling of the topics covered by the researchers working inthe organization and then place them onto the hierarchical tree. Eachof the steps is performed by using our original algorithms: one-byonespectral-additive fuzzy clustering SAF and a recursive algorithmPARL for mapping fuzzy clusters to higher ranks of the taxonomy.The latter minimizes the weighted sum of penalties for the chosen"head subjects", and corresponding "gaps" and "offshoots".