Adopting ontologies for multisource identity resolution

Identity resolution aims at identifying the newly presented facts and linking them to their previous mentions. Our main hypothesis is that variations of one and the same fact can be recognised, duplications removed and their aggregation actually increases the correctness of fact extraction. Our approach to the identity problem has been implemented as Identity Resolution Framework (IdRF). The framework provides a general solution identifying known and new facts in specific domains, and it can be used in different applications for processing of different types of entity. It uses an ontology for internal and resulting knowledge representational formalism. The ontology not only contains the representation of the domain, but also known entities and properties. Apart from extracting information from textual sources, we also exploit structured information available in databases mapping the database schema to the ontology and populating the ontology with existing knowledge. Our main goal is not to advocate one criterion among the others, but to introduce widely applicable solution of the identity resolution problem, we present a set of customisable criteria as well as a mechanism new criteria to be added. We have carried two series of experiments in two different business intelligence domains - company profiling and recruitment - achieving rather encouraging result.