Privacy-driven design of Learning Analytics applications – exploring the design space of solutions for data sharing and interoperability

Privacy-driven design of Learning Analytics applications – exploring the design space of

solutions for data sharing and interoperability

Tore Hoel and Weiqin Chen

Oslo and Akershus University College of Applied Sciences

Norway

EP4LA@LAK15 workshopPoughkeepsie, NY - March 16 2015

The line of argument in this paper

• Privacy, control of data, and trust are essential to implementation of LA solutions

• What does it mean to give priority to those issues?

• Privacy-by-Design principles are «written into» Data Protection Regulations, but what does it mean in design of new solutions?

• Privacy-driven design is good, but we need tool support to make it a reality for learning analytics

• Therefore, the Learning Analytics Design Space (LADE) model

2

The challenge

3

Have LA setbacks implications for design?

4

Too ambitious?

Too big?

How is trust built?

Privacy in which context?

Smaller solution more viable?

5http://www.wyversolutions.co.uk/cms/2015/01/31/xapi-barcamp-at-learning-technologies-2015/

Give students (and parents) ownership of their own data

6

Understanding Privacy

7

First, the LAK community must address privacy!

8

• LAK14 papers:

• Privacy recognised, but only superficially so

• Privacy mostly seen as a barrier

• Privacy hardly defined

Privacy about Limitation and Control?

“debate regarding privacy has swung between arguments for and against a particular approach with the limitation theory and control theory dominating” (Heath, 2014)

9

Respect for Context

• Contextual Integrity:

• «when we find people reacting with surprise, annoyance, indignation, and protest that their privacy has been compromised, we will find that informational norms have been contravened, that contextual integrity has been violated» (Nissenbaum, 2014)

• Informational norm

• Actors, information types, transmission principles

• Contexts:

• Technology

• Business model or practice

• Sector or industry

• Social domain 10

The LADS model

11

12

The Problem and Solution Spaces

13

Design Space

• Questions: Key issues structuring the space of alternatives

• Options: Possible alternative answers

• Criteria: Basis for evaluating and choosing

14

15

Validation of the LADS model

16

Case study – following the data

17

CEDS conceptual model

The Problem Space

• Context of formal study or teaching is essential as it establishes the boundary for what is within or outside the scope of data available for learning analytics

• Socio-cultural barriers have more weight than technical or legal barriers (even if a solution has to involve all three)

18

Solution Space

• Technical: Design of a specification allowing user to express detailed conditions for data sharing when signing in to LA applications, with opt-out possibilities

• Socio-cultural: To boost trust in LA systems, development of privacy declarations, industry labels guaranteeing adherence to privacy standards, and other means of supporting customer dialogue about privacy

• Legal: Ownership and control of data from learning activities are strengthened in national and international law

19

Design Space

20

Criterion:Does the proposed option pass the test of having

been subject to an informed public deliberation on the benefits of LA and the consequences of data sharing for the user as well as for the institution,

the service provider, etc?

Conclusion

• An iterative process model with 3 sub processes (Problem, Solution, and Design) could contribute to a better conversation about design of learning analytics applications

• Charged with a Privacy-by-Design perspective the first round of development and evaluation of this model results in the following recommendations:

• Context integrity may be easier to maintain with smaller LA solutions (limit the scope)

• Socio-cultural aspects of negotiating access to data should direct design of technical an legal solutions

• Learners and institutions need to negotiate the boundaries between informal and formal learning, and institutionally provided tools and technology for personal use

21

Hoel, T. (2015). «Privacy-driven design of Learning Analytics applications – exploring the design space of solutions for data sharing and interoperability » – paper presentation at EP4LA workshop at LAK15, Poughkeepsie, NY, USA, March 16, 2015

@toreabout.me/[email protected]

This work was undertaken as part of the LACE Project, supported by the European Commission Seventh Framework Programme, grant 619424.

These slides are provided under the Creative Commons Attribution Licence: http://creativecommons.org/licenses/by/4.0/. Some images used may have different licence terms.

www.laceproject.eu@laceproject

22

Participate in our study onData Sharing barriers and enablers

bit.ly/lashare

23