Future Enhancements for the Ontology
Although instances and classes were successfully imported to the ontology, the external files worked with were not in RDF or similar format, and therefore, entities from them were given their own namespaces. Future versions of the ontology could be more connected to existing ontologies (in the current version of ontology, there are only one or a couple of such external links from existing ontologies [besides basic elements like rdf:, rdfs: and owl:] , such as 'foaf:knows').
Furthermore, although the ontology contains a detailed branch for scientific fields, the instances are not yet mapped to these fields (though instances are linked to topics) . Mapping topics (i.e., publications) to fields of science is likely to be a month or two-long project on its own, and in order to facilitate insights about interdisciplinary research, a very necessary step in the immediate future of the project.
Inferencing
The ontology used plenty of class restrictions and has been able to make meaningful inferences on the imported VU-Pure data. For instance, fig. \ref{994928} shows imported articles serving as a clue for inferring which journal they belong to. In the earlier versions of the ontology, there were, in fact, more inferences being made due to somewhat more liberal class definitions being in use (fig. \ref{371831}, also see fig. \ref{797019}). As the external data tuned better and better to the ontology with the development of the Python scripts used to prepare them, the bibliographical information imported also became more detailed over time, and general inferences (e.g., "all things that has an author is a document") were replaced by more precise assertions that came with the imported file (e.g., "this_instance has type article"). In future, more experimentation could be made to increase the number of inferences, although well-prepared and pre-aligned files (e.g., by adding class equivalency information in Python during data parsing, rather than doing this in Protege) may, once again reduce the need for inferencing for crucial information in future work as well. And unfortunately, as mentioned before the prototype does not yet use the scientific field class (i.e., it is not yet specified what article belongs to which scientific field) or organization class (i.e., institution) in inferences . These (crucial) features will be implemented after completion of the course as part of the research project, and inference will likely play an important role in inferring scientific fields from the topics (i.e., keywords) of articles.