COMA++: Results for the Ontology Alignment Contest OAEI 2006

This paper summarizes the OAEI Contest 2006 results for the matching tool COMA++. The study shows that a generic schema matching system can also effectively solve complex ontology matching tasks. 1 Presentation of the system COMA++ is an extension of our previous COMA prototype [1]. It is a customizable and generic tool for matching both schemas and ontologies specified in languages such as SQL, XML Schema or OWL [2]. COMA++ offers a GUI and supports the combined use of several match algorithms as well as the reuse of previously confirmed match results [6]. The COMA++ architecture is shown in figure 1. The Repository persistently stores all match-related data, the Model and Mapping Pools manage all schemas, ontologies, and mappings in memory, and the Matching Engine performs the match operations. The GUI provides access to these components and is used to visualize models, manage the match process and mappings. The Matching Engine contains different libraries that supports many match algorithms and match strategies. The similarity results of individual matchers are maintained and aggregated within a similarity matrix per match task [1]. Match strategies implement workflows to deal with complex match tasks and enable a reuse of previous results and the decomposition of larger match tasks into smaller ones [3]. 1.1 State, purpose, general statement COMA and COMA++ have proven to be very effective for matching database and XML schemas [1, 4, 6]. The main reason for this test was to see the effectiveness of a generic matching tool for dealing with ontologies. Repository Graphical User Interface Matching Engine (Fragment-based) Model Pool External Schemas, Ontologies Mapping Pool Exported Mappings Resolution Library Matcher Library Combination Library Model Manipulation Mapping Manipulation So urce Id Name Struc ture Content SOURCE O bje ct Re lId So urce Re lId O bje ct1 Id O bje ct2 Id Ev ide n ce O BJECT_ REL