Document Type


Publication Date



Relationships play a key role in Semantic Web to connect the dots between entities (concepts or instances) in a way that enables to absorb the real sense of the entities. Some interesting relationships would give proof for the existence of subject and object in triples which in tern can be defined as evidential relationships. Identifying evidential relationships will yield solutions to some existing inference problems and open doors for new applications and research. Part_of relationships are identified as a special kind of an evidential relationship out of membership, causality and etc. Linked Open data as a global data space would provide a good platform to explore these relationships and solve interesting inference problems. But this is not trivial because LOD does not have a rich schema in terms of the data sets and also the existing work with respect to schema mapping in LOD is limited to concepts and not relationships. This project is based on finding a novel approach to identify partial relationships which is the superset of part_of relationships from LOD instance data by conducting a proper analysis of the data patterns in instance data. Ultimately this approach would provide a way to enhance the shallow schemas in LOD which in tern would be helpful in schema matching in LOD. We apply the determined approach to the DBpedia data set in order to identify the partial relationships in DBpedia.


Kno.e.sis Research Center Technical Report