GOMMA results for OAEI 2012

We present the OAEI 2012 evaluation results for the matching system GOMMA developed at the University of Leipzig. The original application focus of GOMMA has been the life science domain but as a generic tool it can also match ontologies from other areas. It could thus participate in all OAEI tracks running on the SEALS platform. GOMMA supports several methods for efficiently matching large ontologies in particular parallel matching on multiple cores or machines, reducing the search space as well as reusing and composing previous mappings to related ontologies. 1 Presentation of the system 1.1 State, purpose, general statement GOMMA (Generic Ontology Matching and Mapping Management) [6] is a comprehensive infrastructure to manage and analyze the evolution of life science ontologies and mappings [4]. It includes a generic component to semantically align (match) ontologies. GOMMA is able to match very large ontologies as common in the life sciences. To deal with large ontologies GOMMA provides several scalable match techniques: 1. Parallel ontology matching on multiple computing nodes and CPU cores [2], 2. Indirect computation of ontology mappings by reusing and composing previously determined ontology mappings via intermediate ontologies [3], and 3. A newly introduced blocking approach to reduce the search space by restricting matching to overlapping ontology parts. These techniques all support efficiency, in particular reduced computation times. The latter two approaches can also improve match quality. While the original focus of GOMMA has been in the life science domain, the match component is generic. We could thus participate in all 2012 match problems of the Ontology Alignment Evaluation Initiative (OAEI)1 running on the SEALS platform. 1.2 GOMMA Matching Workflow The GOMMA matching workflow used for OAEI 2012 is displayed in Fig. 1. In the following, we describe its three main phases, namely the initial phase (including the new blocking strategy), the matching phase as well as a set of postprocessing steps. 1 http://oaei.ontologymatching.org