Ontology Mapping: A concept hub for storing mapping data and linguistic methods for producing mappings

Mapping between Knowledge Organization Systems (KOS), generally called Ontology Mapping for short (ontologies being a type of KOS) is central to achieving interoperability between systems, such as Electronic Health Record systems, and to reaching the full promise of the Semantic Web by finding any answer that can be derived by combining data posted by many agents. The problem of ontology mapping and its application can be divided into (1) designing systems that hold mapping data for use and (2) methods for creating mapping data. I will approach both issues from a deep conceptual perspective. On (1), mapping data are often stored "bilaterally" for individual pairs of KOS. I propose an alternative, dubbed concept hub, which takes the approach used in the Unified Medical Language System (UMLS) one step further. In a concept hub for a group of KOS, concepts from each KOS are stored as description logic formulas, using a base vocabulary of elemental concepts and of concept relationships. The relationship between any pair of concepts in the hub can then be computed by a reasoner. On (2) I will present an approach for mapping from complex terms for the material of an art object, such as acrylic paint on canvass mounted on wood, as found in records in the ARTstor database, through elemental concepts from the Getty Art and Architecture Thesaurus (AAT). In the example, the description logic formula is surface canvas; subsurface wood; coating acrylic paint. The mapping process can be partially automated by discerning linguistic patterns that correspond to description logic patterns, with the role of each element in the description logic pattern can be seen from the place of a term in the linguistic pattern.