A Method to Transform Datasets into Knowledge Graphs

Knowledge graphs are representations of data and information about resources in a triplet-based format which are identifiable by unique IRIs, are reference enabled and expansible; these characteristics make knowledge graphs easy to upload and manage large volumes of data in an agile way. In this article we propose a semi-automatic method for transforming datasets into knowledge graphs. Specifically, we describe the method in the transformation of a set of files representing the logs of a medical research protocol whose purpose is to evaluate the efficacy of the use of continuous glucose monitors in patients with Type 1 diabetes. For evaluation purposes we implemented a set of programs to perform data extraction from the dataset, parsing, cleaning and finally the automatic population of the knowledge graph. The resulting graph has been evaluated by verifying its logical consistency.