Towards ontology matching maturity: contributions to complex, holistic and foundational ontology matching

Interoperability between different ontologies and their instances is at the core of the Semantic Web. Ontology matching and instance matching, as distinct tasks, aim at facilitating the interoperability between different knowledge bases at their terminological and assertional levels, respectively. Both tasks are key ones in many applications as they are the basis for data exchange and integration. Ontology matching, in particular, is the task of generating a set of correspondences (i.e., an alignment) between the entities of different ontologies. This is an active research area that has fully developed over the last two decades, with approaches being systematically evaluated in the context of the Ontology Alignment Evaluation Initiative (OAEI) campaigns. A variety of matching approaches have been proposed so far, mainly dealing with the generation of simple alignments (i.e., alignments involving singleentities from each of the ontologies to be matched) between pairs of ontologies (i.e., pairwise matching). Those approaches have as well been mainly applied on ontologies with the same level of abstraction, in particular domain ontologies (i.e., ontologies describing the entities related to a particular domain). However, simple correspondences are proven to be insufficient to fully cover the different types of heterogeneity between ontologies. More expressive (complex) correspondences are required instead. Moreover, in domains where several ontologies describing different but related aspects of the domain have to be linked together, matching multipleontologies (i.e., holistic matching) simultaneously is necessary. Last but not least, while linking domain ontologies to foundational ontologies has proved to improveontology quality and interoperability between domain ontologies, the problem of matching this kind of ontologies has been addressed to a lesser extent in the field.This is a challenging task, specially due to the different levels of abstraction ofthese ontologies. These matching contexts raise as well the problem of automatically evaluating the corresponding approaches, which requires dedicated evaluationstrategies. This manuscript presents my contributions to the ontology matchingfield, in particular on the generation of expressive (complex) alignments, matchingof multiple ontologies and foundational and domain ontology matching. It is thesupport of my habilitation to supervise research