Abstract:

Evaluating and analysing learning-process-data tracked from learning experiences (e.g. made in a WBT) is a common task in the field of Learning Analytics. For more complex analyses there often emerges a need for detailed information about the content that is used for learning and about the structure of the whole learning-unit so that extensive, detailed knowledge about all used Learning Objects is needed. Currently, one of the most commonly used standards for tracking learning-process-data is the Experience API (xAPI). Deplorably the methods provided by the xAPI to store and manage information/metadata about the learning-content itself are defined partially insufficient and inefficiently.

In the paper we discuss in detail what problems may occur when the xAPI-standard is strictly implemented in Learning Analytics projects. To surpass some limitations of the xAPI specification we develop the concept of the “Learning Activity Provider” (LAP). The LAP is a webservice that resides inside or alongside a Learning Record Store (LRS) and is used to store, manage and provide Learning Object definitions in a format that satisfies the xAPI specification. Therefore, the LAP leverages the concept of a “Metadata Provider” (as defined in the xAPI specification [1]) to a more powerful system that can efficiently manage xAPI-driven Learning Objects in real-world Learning Analytics contexts.

We introduce the fundamental concept of the LAP and how it can be integrated in existing xAPI implementations as well as a basic set of functionalities that were derived from different concepts of Learning Object Repositories and the Metadata Provider. Furthermore, we propose suitable Learning Object and metadata models which are based on an extensive research of common Learning Object and metadata models [2] as well as the Activity and Activity Definition models, introduced by the xAPI [1]. Because manually adding and managing metadata is a tedious and time-consuming work, we propose ways to generate them full- or semi-automatically. Furthermore we suggest to connect the LAP to the LRS that stores the tracked learning-process-data to extend the metadata model with dynamically created, additional paradata [3]. In conclusion, we recap the new possibilities that an implementation of the LAP would offer in a xAPI-driven Learning Analytics process.

References:
[1] xAPI Specification - Activity Data and Metadata, Accessed 21 November, 2019. Retrieved from https://github.com/adlnet/xAPI-Spec/blob/master/xAPI-About.md#activity-data-and-metadata
[2] S. Yassine, S. Kadry, and M. A. Sicilia, “Learning Analytics and Learning Objects Repositories: Overview and Future Directions,” in Learning, Design, and Technology, M. J. Spector, B. B. Lockee, and M. D. Childress, Eds. Cham: Springer International Publishing, pp. 1–30, 2017
[3] West, Brady T. "Paradata in survey research." Survey Practice 4.4, 2011

Keywords:

Learning Analytics, xAPI, Learning Objects, Metadata, Paradata.