Representing scientific experiments: implications for ontology design and knowledge sharing

As part of the development of knowledge sharing technology, it is necessary to consider a variety of domains and tasks in order to ensure that the shared framework is widely applicable. This paper describes an ontology design project in experimental molecular biology, focusing on extensions to previous ontological models and frame-based formalisms that allow us to handle problems in the representation of experimental science knowledge. We define object histories, which are used to track substances through a series of experimental processes, including those which transform their participants from one category to another. We define object and process complexestemporary configurations with features of their own. We present extensions to a frame-based formalism to support these features. Additional features of our frame formalism include slot groups for identifying sets of relations with common properties, and partial filler restrictions that combine knowledge of the most likely slot values with the ability to handle unexpected values. We demonstrate how these extensions enable the use of (relatively) domain independent inference rules, support intelligent information retrieval, and improve the quality of query interfaces; and we describe the translation of our formalism into Ontolingua.