Information integration arises nowadays as a main problem in different application scenarios like datawarehouses, information retrieval and web portals, e-business and virtual enterprises. Mainstream initiatives in these domains are exploiting the concept of ontology, understood as “a formal, partial and explicit specification of a shared conceptualization”. Ontology is a key concept in knowledge based systems in general and in Semantic Web in particular.
However, interoperability actors do not always agree on the information being shared, justifying the use of distinct ontologies, even if corresponding to the same domain of application. To overcome ontology-based interoperability clashes, ontology mapping proved to be an efficient solution. However, the ontology mapping automation is very difficult and error-prone due to the subjectivity of the problem.
MAFRA Toolkit is an ontology mapping tool developed by the research team proposing this project, which provides limitedly automatic semantic bridging support. Due to their service-oriented architecture, MAFRA Toolkit relays on Services many of the competences otherwise centralized into a whole big monolithic rule-based system.
Despite the advantages arising from this approach, too many semantic relations are often suggested for a group of entities. These are clearly situations of ambiguity, in the sense that too many Services are triggered due to too few specified constraints.
This project aims to reduce the ambiguity arising from the similarity measures provided by the similarity measuring algorithms already applied on MAFRA, by researching on the potentialities of the information arising from the formal methods applied in ontology engineering processes, such the ontology assemble process and pattern-based ontology engineering. Our rationale grounds on the formal nature of these processes, especially those developed in context of the FONTE research. In this research, a set of formal, univocal ontological relations and axioms between ontologies being assembled have been developed and successfully adopted. Other ontology engineering processes, such pattern-based development, Methontology and OntoClean are also expected to provide useful information and should therefore be analyzed too.
We aim to analyze and systematize the relevance of the information resulting from these processes, and develop a set of Matches capable to exploit such information into the automatic bridging process. This type of information has not ever been used in this context, and is our conviction that it will provide efficient disambiguation evidences which will motivate major benefices for the application domains.