Abstract |
There is a relatively high amount of structured knowledge available in the field of biomedicine. For example, the BioPortal repository provides access to 327 ontologies with a total of 3M links between their concepts. Where a lot of research has gone into evaluation of the quality of the links, we set out to assess the quantity of links, tackling the question "How linked is this data?" First, we apply network analysis: each ontology is a node, and the number of links between them gives us weighted edges. Second, we look at this question from the perspective of a particular usage scenario. In the EURECA project we have several datasets that we need to integrate. They are annotated with different bioportal ontologies. To what degree are the existing links between bioportal ontologies enough to accomplish this goal? We show that the assessment of the availability of links is much more positive when we take into account the usage of the concepts in the ontologies. |