Konstantin Todorov                                                                                                                               

Contact - PhD Project - CV - Conferences & Publications - Teaching - Other

PhD Project:        Ontology Matching by Combining Structural and Extensional Similarities                                                                            

Supervisors:         Kai-Uwe Kühnberger and Peter Geibel

Abstract:             Ontologies are knowledge bodies describing the semantics of data and are broadly applied in supporting knowledge exchange between different parties. However, the adoption of the same ontology over a certain domain of interest by different people or organizations is unlikely, mainly due to the decentralized and strongly human-biased nature of ontology development. For that reason one needs to dispose with a procedure of mapping the elements of two distinct ontologies that are likely to share a significant overlap of the domains that they describe in order to enable interoperability between agents.

The project proposes a solution for mapping hierarchical ontologies, populated with properly classified text documents, used as concepts instances. It combines structural and instance-based similarity approaches in order to yield concept-to-concept mapping assertions between two source ontologies, as well as overall similarity, granularity, instantiation and specificity judgments. The solution is accomplished by applying machine learning, natural language processing and graph matching techniques. It can be successfully applied to mapping web directories and will be further generalized for non hierarchical structures.

Areas of research interest

Ontologies and Ontology Matching

Machine Learning, Support Vector Machines, Feature Selection

Natural Language Processing, Text and Data Mining