Big Data

Learning Analytics in Informal Scenarios

Dr. Stefan DietzeHannover (GER), Dezember 2016 - Learning is increasingly taking place in informal contexts and is a central component of everyday online activities. For example, the exchange and interaction in social networks and the use of search engines often reveal an implicit learning process. This produces large amounts of data on a daily basis whose interpretation in relation to factors of relevance to learning, such as the user’s competence or learning behavior, requires efficient evaluation. Dr. Stefan Dietze of Hannover’s Leibniz University will provide details in his lecture at LEARNTEC.


Your LEARNTEC lecture is entitled "Analytics for Everyday Learning" (AFEL). To what extent do analytics in the daily learning process differ from those used for specific, defined learning activities?

Dr. Stefan Dietze: Learning is increasingly taking place in informal contexts and as an aspect of everyday online activities, e.g. as part of our interactions on Facebook, Wikipedia, and Slideshare or as an implicit element of web searches, in which learning processes can be recognized. This is in stark contrast to the tightly connected defined limits of formal learning scenarios and environments in which learning goals that are usually defined are supported by dedicated learning resources. This also poses new challenges in the realm of analytics.

There are still essential issues, including how learning activities can be identified in such informal scenarios and how learning success or learning-relevant concepts such as competence can be recognized from large, heterogeneous activity data sets. A central problem, among other things, is the absence of assessment data that provide information about learning success. Learning analytics therefore has to include heterogeneous and extensive data, e.g. from activity streams, navigation data, and "behavioral traces" such as mouse movements, in order to draw meaningful conclusions.


How does the use of learning analytics influence and change the learning process?

Dr. Stefan Dietze: One approach is, e.g. to use conclusions gained in regard to learning behaviors in the social web in order to provide better learning support through appropriate help and recommendations, or early recognition of negative behavior in order to be able to react appropriately. As part of the AFEL project, initial results are being applied in the context of the Didactalia social environment, with other platforms to be added later.

What extra effort - technical and otherwise - does the deployment of learning analytics mean for education providers?

Dr. Stefan Dietze: Informal environments in particular require greater technological familiarity for the analysis of heterogeneous data sets. From the analysis side, machine learning techniques are of great importance, as are general data-science methodologies. Furthermore, due to the complexity and quantity of the (historical) data involved, technologies for efficient distributed processing, i.e. the methods that generally fall under the rubric of "big data" technologies, are usually helpful. This, however, this is not specific to learning analytics, but applies to all data-driven scenarios. Learning analytics is just one of the many areas of application that can lead to better understanding of user activity.

What are the benefits for the providers and the users?

Dr. Stefan Dietze: Both providers and users benefit primarily from having better understanding of the users in that better support can be made available for them, their tasks, and their information needs.

When do you believe the use of learning analytics will become standard procedure?

Dr. Stefan Dietze: Depending on the definition, learning analytics can already be regarded as a standard practice. In the AFEL approach, especially in informal scenarios, e.g. in the social web, user activities are already extensively analyzed and examined. In more formal learning scenarios, data mining and learning analytics have long been an implicit and omnipresent tool.

"Analytics for Everyday Learning - Mining & Analysis of Learning Activities on the Web", Conference Room 8/9, 24 Jan 2017, 11:45-12:15