III-COR-Small: Efficient Matching for Large Real-World Schemas and Ontologies
PI: Maria Cruz
This project aims at bridging across heterogeneous data and will impact applications in multiple fields, including emergency management, biomedicine, digital government, and environment. In particular, this project extends the state of the art in schema and ontology matching, and therefore in data integration, by testing and evaluating methods and strategies that establish relationships among semantically related concepts in heterogeneous data sources.
The following research issues are addressed: (1) Design of methods and algorithms for schema and ontology matching that operate at different levels of granularity (e.g., concept, structure); (2) Development of a prototype of an integrated system that supports the visualization and manipulation of large schemas and ontologies in addition to the developed matching methods and algorithms; (3) Test and evaluation of the above methods and system prototype in terms of their effectiveness, including accuracy (precision, recall) and efficiency (execution time). From an educational viewpoint, this project impacts the design of courses and the research training of graduate and undergraduate students, and of a postdoc.
Further information about this project can be found at: www.cs.uic.edu/~ifc/grants/SchemaOntologyMatching/
The research work under this award followed the objectives defined in the abstract of the project. In particular, our work aims at integrating heterogeneous data with a focus on schema and ontology matching. The following research issues have been addressed: (1) Design of methods and algorithms for schema and ontology matching that operate at different levels of granularity (e.g., concept, structure); (2) Development of a prototype of an integrated system that supports the visualization and manipulation of large schemas and ontologies; (3) Testing and evaluation of matching algorithms in terms of their effectiveness, including accuracy (precision, recall) and efficiency (execution time). We have collaborated with experts in the biomedical and geospatial domains and AgreementMaker has been used by more than one hundred institutions around the world, whereas AgreementMakerLight, which has resulted from our collaboration with the University of Lisbon, is available open source. AgreementMaker and AgreementMakerLight are among the top systems in the world. Recently, AgreementMakerLight has won six tracks in the Ontology Alignment Evaluation Initiative (OAEI) competition (October 2014). From an educational viewpoint, this award has impacted the design of courses and the research training of graduate and undergraduate students, and of postdocs. We have published in the main conferences in databases and semantic web, namely VLDB, ICDE, ISWC, ESWC, and EKAW. According to Google Scholar, our 2009 VLDB paper "AgreementMaker: Efficient Matching for Large Real-World Schemas and Ontologies" has been cited 132 times (as of November 30, 2014) including by the most authoritative surveys of the field.
