Data, Data Formats and Learning Analytics Infrastructures
In our group and in some of our projects, e.g. Open Digital Lab for You (DigiLab4U - https://digilab4u.com/) we decided to work with one of the de facto standards for learning data collection after conducting comprehensive research: xAPI (https://xapi.com/overview/). This data format supports many contexts, is flexible, supports multi-modal applications and many technologies that we use in our project. Especially exciting is that one of the features is that we can merge learning data from different systems in our Learning Record Store to get deeper insights into the learning process.
Another advantage of the specification is that the basic idea is simple. Every statement we store consists of at least three components: an actor, a verb, an object. The actor is the learner, which is linked to the statement. She is identifiable via email or other data. The verbs explain, what the learner did. The object typically links the statement to an activity, which is linked to the verb. In addition to that, it is possible to use different kinds of extensions to specify the context.
Here are two examples with different complexity:
- John Doe downloaded the laboratory hand-out.
- Jane Doe completed the measuring chamber course with a passing score of 98%.
Sounds great? Yes, but the work with this kind of data format has some pitfalls. In some cases, there are different possibilities on how to express a statement, because the difference between verbs and objects can be fuzzy. The solution in the specification are self-explaining identifiers, URIs, which help us to understand and interpret the collected data. Furthermore, metadata is often missing in the common collections, contexts are missing and there is no possibility to see if there were changes in the metadata.
Because we want to create FAIR data in our project (Findable, Accessible, Interoperable and Reusable), we are working on a concept to improve or expand xAPI. One way to do this, is to discuss this with our project partners, e.g. the consortium in DigiLab4U provides insights, ideas, needs and research questions, especially for lab-based learning. To create an even better concept, we are also working with many other partners, who help us to create a solution, which is able to last (sketch below). Our efforts to create sustainable definitions fo xAPI data are not necessarily new, but there is still no comprehensive solution.
We are conducting a study that will contribute to standardization, generalization and a scientific consensus. We want to initiate a common, collaborative registry. To ensure our goals and avoid an isolated solution, we depend on more help.
There are different opportunities for cooperation with us:
- As owner/maintainer for a shared repository of definitions
- Contribute definitions as a developer
- Co-development of the frontend
- Discussion on (Process|Concept|Contents|Infrastructure)
Please feel free to contact Matthias Ehlenz or Birte Heinemann, if you are interested.