Domain Modeling

Domain

In order to explore the effects of interdisciplinarity and discover scientific collaboration patterns, a domain model was created, which consists of properties and resources that are related to scientific publications, such as type of publication (e.g., journal, book), authors, topics, scientific fields, and so on. Since scientific collaborations of today's academia generally result in some sort of publication, and there are ample records regarding to these publications, they are highly likely to be good probes to measure of research and collaboration patterns. Therefore, modelling the landscape of scientific publications via an ontology, and populating it with instances from various datasets is viewed as a valid approach.
The instances for the ontology is currently drawn from a bibliometric database, Elsevier's Pure \cite{pure}, and more instances will be later added from RISIS \cite{20172017}, and Web Of Science \cite{analytics2017}. In interdiciplinarity research, like other meta-scientific research topics, this bibliometric approach is an often used and considered an effective method \cite{Roessner_2012,Mugabushaka_2016,Zulueta_1999,Perianes_Rodriguez_2016,Cardona}.  The methodology used in the current project is detailed further in the proceeding section.

Methodology and Ontology Building Process

Ontology Creation and Revisions

In order to create a model of the domain, a first prototype of an ontology was developed during the past month as part of the course, and this model was revised and improved through project meetings.  As the ontology progressed —and I gained more experience— methodological changes occurred and the ontology was significantly changed between revisions. Most notably, the range and class restrictions that were often applied with 'rdfs:range' and 'rdfs:domain' properties were entirely removed due to the reasoning errors and inflexibility they lead to, and also as per expert recommendations (now, more delicate range and domain restrictions are applied through equivalency and subclass relationships where needed). The latest version of the ontology features more sophisticated class definitions (see Fig. \ref{490596}) through equivalency statements and this results in a more stable ontology and more reliable inferences. Besides the technical advancements, the structure of the ontology was updated based on project meetings. Therefore, the current version consists of a more comprehensive, accurate, and stable model of the domain.