By Tori Schmitt
The purpose of creating and using an ontology within the context of a database is to discover new knowledge. As we learned in the first Musiconis session on ontologies and linked data, a well-constructed ontology allows for the information contained to be found quickly and with ease through systematic through systematic processes, one of which can be defined as ontological reasoning. In Musiconis’ second session on ontologies (January 12, 2017), Dr. Victoria Eyharabide further explained the processes behind this concept and provided hands-on instruction (in the program Protége) bridging theoretical planning and praxis.
Ontological reasoning is a process which improves database search ability by linking individual entries within the database to equivalencies and inferences through the function of the database reasoner. When making an inference, one is making a logical conclusion based on a logical premise. Within the database it is the same process and inferences are made both manually, by the person using the database, and by the built-in database reasoner (in the case of the program Protégé, the reasoner is called HermiT.) In doing so, individual entries are improved through “inference” and become linked to more than just the specific metadata attributed in their record file. Subsequently, expanded metadata improves search functions. An example of this would be an image that is indexed as “David plays the harp.” Without database inference, the entry would only be found using the search terms “David” and “plays” and “harp.” However, using the database reasoner the entry would also be linked with the broader categories of its classes and object properties. In this example, the reasoner, on the most basic level, would link the entry with “person” and “chordophone,” the two broader classes which contain “David” and “harp,” allowing the entry to be found under a broader umbrella of search terms. This type of inference is referred to as subsumption checking. Other types of inference made by the reasoner are equivalence checking (if class A is equivalent to class B and class B is equivalent to C then A is equivalent to C), and consistency checking (looking for errors). Overall, the use of the reasoner helps to establish that the ontology is working logically and following all established rules.
Another key aspect of ontological reasoning is the world assumption. A closed-world assumption is how most databases and computer systems are coded. It operates on a strict binary and excludes all instances that are not explicitly given. In contrast, an open-world assumption, which the Musiconis database uses, operates on the notion that the data is always accurate but incomplete. This manifests in an indexing of information as yes or we don’t know, rather than a simple yes or no. Through this, open-world assumptions allow for indexing of uncertainty- a function crucial for accurately indexing the present knowledge state of medieval musical iconography. While it is easy to assume that working with database ontologies is technical and separate from the field of humanities study, it was clear in Dr. Eyharabide’s session that here, in the Musiconis database, technical praxis is intended to express our understanding as medievalists and that in order to create a successful ontology of any discipline, one cannot separate the project’s parent methodology from its ontological reasoning.