Assessing applied ontologies by distinguishing formal and material content
Kohl, Ryan Conrad
MetadataShow full item record
According to the editors-in-chief of the Applied Ontology journal, the field of Applied Ontology studies the ways in which models of the world can be created and how they can be used with data models and knowledge representations to add value to information systems (Guarino & Musen, Applied Ontology: Focusing on content, 2005)[p 1]. There are plenty of benefits to be had by creating these world-representing models, from a clearer understanding of the 'meaning' of data to an easier way to share data. These benefits are matched by many problems that arise in the process of doing applied ontology. One commonplace problem is deciding whether an ontology is a good model of the world. Particularly when two ontologies are compared and contrasted, people often run into difficulties the basis on which to draw such conclusions. Of course a good ontology needs to be logically consistent, so that contradictions don't arise from the applications of its rules of deduction. But more than this, we would want to define a good ontology to be sound, meaning that only true statements can be derived from it, given that only true premises are asserted. In a sense, a sound ontology would not have any 'false rules'. Generalizing this, a good ontology would have neither false assertions nor false rules, guaranteeing that only true statements were derivable. The problem is that it is easy to end up disagreeing whether a statement is true. Almost a century ago, the logical positivists were discussing the same problem. If a statement is true when it accurately corresponds to the world, how can two people resolve a disagreement over this correspondence? In the early 1930's, philosophers such as Carl Hempel and Rudolf Carnap decided to invest certain statements - "protocol statements" - with a particular importance: these statements were intended to be taken as true by reason of being asserted by a trained scientist who had taken the proper precautions to guard against error. The definition of an ontology (an anachronistic name for the scientific logical systems they were discussing) was altered to allow for differing protocol statements, enabling a large group of people to make assertions and draw acceptable conclusions from the same system while maintaining their disagreements with one another. Moritz Schlick disagreed with this alteration and, in a series of papers with Hempel, argued that the distinction between formal and material statements in an ontology (again, an anachronistic name) was essential in understanding what was being disagreed on. By making this distinction, Schlick ended up supporting a very different use of ontologies that encouraged the creation of semantically rich yet idiosyncratic models of reality. Ontologies were composed of formal statements that provided instructions on how to describe the world and material statements that (attempted to) be descriptions of the world. In this thesis, we will review the debate over protocol sentences and reference between Carl Hempel and Moritz Schlick, concluding with Schlick's definitions of formal and material statements. We will then build on his work by defining formal and material terms, which are intended to aid applied ontologists when reviewing and building ontologies. Finally, we will review the Ontoclean methodology of ontology creation, concluding that a failure to recognize the difference between formal and material ontology content can lead to needless complications in ontology construction.