Exploring MOOC metadata - Data available via the MOOChub format

With more and more learning material available online, it became necessary to support learners in finding appropriate learning resources (e.g. courses). Metadata plays an important role in this task because it is an easy-to-process and often machine-readable data source for designing filters, search engines, and (AI-driven) learning path assistants. The quality of these services strongly depends on the availability of metadata. This study sheds light on the metadata supplied by a group of MOOC providers organized in a consortium supporting the same metadata format as a de facto standard. We focused on what data is provided and if it fits the needs of metadata consumers building different services processing these data. Our approach was to compare the supported and provided attributes according to the standard. The data was provided as a JSON file, which can be easily investigated by e.g. Python scripts. The JSON file can be treated as a dictionary with key-value pairs. The keys represent the attributes while the corresponding value is the attribute value. We compared which optional attributes were supported (the key exists in at least one of the courses) by the different providers and how frequently they were supplied (the value was not None/null or a placeholder like an empty string). The paper at hand gives insights into what metadata the course providers deliver and made it possible to evaluate the usefulness of the metadata format.