InterOnto - Ranking Inter-Ontology Links

Entries in biomolecular databases are often annotated with concepts from different ontologies and thereby establish links between pairs of concepts. Such links may reveal meaningful relationships between linked concepts, however they could as well relate concepts by chance. In this work we present InterOnto, a methodology that allows us to rank concept pairs to identify the most meaningful associations. The novelty of our approach compared to previous works is that we take the entire structure of the involved ontologies into account. This way, our method even finds links that are not present in the annotated data, but may be inferred through subsumed concept pairs. We have evaluated our methodology both quantitatively and qualitatively. Using real-life data from TAIR we show that our proposed scoring function is able to identify the most representative concept pairs while preventing overgeneralization. In comparison to prior work our method generally yields rankings of equivalent or better quality.