Larry Lefkowitz, Cycorp, Inc.
Two major causes for optimism in the Semantic Technologies community are: 1) the rapid growth in the availability of open (linkable) structured datasets and associated ontologies, and 2) the belief that shared representation schemata will enable ontology integration and data sharing. Increasing pairwise linkages, as reflected especially at the dataset (instance) level by the growth of the Linked Open Data Cloud, offers evidence to support this optimism.

This apparent smooth sailing arises in part, from the relatively simplicity, and in some cases lack of precision, of the representation of the content and ontologies being merged. While this simplicity is beneficial in making such linkages possible, it can hide some symptoms of underlying problems that would best be revealed and addressed sooner rather than later. Alas, there may be some bumpy skies ahead. Over the past year, Cycorp has made a concerted effort to link the Cyc ontology to a number of external structured datasets and ontologies, taxonomies, and thesauri. This experience has revealed several areas in which knowledge representation issues in many current data sources and ontologies have been addressed non-uniformly, sub-optimally, or not at all. This presentation will examine some of these representational challenges, with an emphasis on their impact on ontology integration as well as a focus on suggestions for representational improvement.