Applying evolutionary terminology auditing to the Gene Ontology.

Evolutionary Terminology Auditing (ETA) is a novel way to assess the quality of terminologies using reality as benchmark. The key idea is that terms added to each new version of a terminology reflect unjustified absences and terms that are deleted unjustified presences in previous versions of the terminology. The method requires that terminology authors not only keep track of changes in successive versions, but also motivate the changes introduced. In this paper, we report on how our method has been applied to the Gene Ontology (GO), a collection of three structured, controlled vocabularies for use in annotating genes, gene products and sequences. We demonstrate that even where the basic requirements for its application are only partially satisfied, the approach can still yield results which are useful for quantifying and forecasting the evolution of a terminology's quality over time.