Tech & Data

This page dates from 2017; we expect to have a similar core infrastructure for 2018 with some new additions…. to be announced.

The following summary of systems and data which will be available for the hackathon is subject to additions, as more people make offers to contribute. If you are interested in bringing some technology or data to the hackathon, please see our Contribute page.

Contributions from Jisc

Jisc is working with a number of UK institutions collecting and using large data sets in a standard format as part of its national infrastructure for learning analytics (pdf).  Core tools and sample data from the project will be made available as part of the hackathon.

This includes:

  • The Learning Records Warehouse, which stores activity data in xAPI and student data in a standard format known as the Universal Data Definitions (UDD). This is based on the HT2 Learning Locker open source product augmented to support Jisc UDD storage.
  • A sample data set including student data, VLE, attendance, and library use as well as sample predictions.  Whilst the data will be generated, it will follow a similar profile to the live data set.
  • Analytics Labs – a secure desktop environment, allowing creation of visualizations using Tableau.
  • Jisc’s Universal xAPI Translator, which allows data to be rapidly converted from text files into any xAPI profile, allowing easy testing of new recipes and profiles.
  • Sample xAPI aggregation code toolkit, showing how to aggregate xAPI and UDD data as the first step in producing visualizations.
  • Access to the API and code for the Jisc student app – Study Goal, allowing experimentation with student facing services.

Lee Baylis outlines some of the Jisc technologies in an informal blog post.

Tribal Activity Data Generator

Tribal has made its prototype activity data generator available as a GitHub repository. This is an early stage development, for use to generate testing and demonstration data, which can simulate various patterns of behaviour for different stereotype student and group types. This has been established with an Open Source licence.

ICTS – University of Amsterdam

synthetic data generator based on open source software that you can use to provision a Learning Record Store or performance/conformance test realistic infrastructure

Related Objectives

The Hackathon has always had an emphasis on “open” – standards, architectures, APIs, data – and we particularly invite participants to contribute some tech/data at any scale which would help us to better aggress the following three enabling objectives:

Student facing Open APIs
While we are increasingly providing LA solutions for students, a significant opportunity arises to investigate the way in which universities and other open government services can enrich and expose their data stores and analytics, thus fostering the rapid development of student facing solutions. Institutions are increasingly moving towards an API based architecture which will add flexibility to their core IT infrastructure, a situation that offers many opportunities for rapid innovation and development of solutions by the people who will use them (rather than by external providers). At the same time, many universities host a pool of highly motivated students with fundamental ideas about how to improve the student experience by offering innovative new IT services that are beyond core business. This challenge is to facilitate such access pathways for systems built using the core infrastructure provided by official university data warehousing situations. We will investigate Security, data format, mapping, and other core properties.

Infrastructure-integrated approaches for the joint exploitation of distributed data sources,
including synthetic data
While previous events have focused on the exploitation of data from single sources, including the synthetic generation of data from those sources and their exploitation in visual analytics methods, a key aspect here regards the integration of the increasing number of data sources. From which we can assess the learning activities, learning environment and learners’ experiences. Here, we will be looking at profiles and technical approaches to enable the joined-up use of data coming as much from institutional systems, as from the learners’ other platforms (e.g. social media) [1]. A specific emphasis here will include the ability to generate synthetic, homogeneous, and exploitable data from the kind of sources manipulated within the AFEL and Jisc projects.

Analytics beyond user-computer interaction data
We will try to move from the datasets of user-tocomputer interactions (the actor-verb-object paradigm) to focus more on the user-to-world interactions which we can elicit with motoric and physiological sensors (the sensor-sample-value paradigm). Multimodal datasets collected from practical and workplace learning experiences will be made available for analysis. These learning experiences are composed of atomic actions defined by an Experience API statement (e.g. the “expert pulls lever”), which will point to a list of multimodal sensor recordings. Some research questions associated with this type arise including: what are the essential features that we can extract from sensor data streams? Which data analysis and techniques are suitable? At the input level, how can we compare two or more action executions. What is the most efficient way to integrate sensor data with xAPI? What is a meaningful visualisation?

References

[1] Bakharia, A., Kitto, K., Pardo, A., Gašević, D., & Dawson, S. (2016, April). Recipe for success: lessons learnt from using xAPI within the connected learning analytics toolkit. In Proceedings of the sixth international conference on learning analytics & knowledge (pp. 378-382). ACM.