Data WG 2022-06-15 Meeting notes

 Date

Jun 14, 2022

 Participants

  • @Edward Zarecor

  • @Dave Ormsbee (Axim)

  • @Jill Vogel

  • @Régis Behmo

  • @John Baldwin

  • @Andrew Shultz

  • @Zia Fazal

  • @Julien Maupetit

  • Sergio Simonian (DATUM/FUN)

  • @Former user (Deleted)

 Goals

  •  

 Discussion topics

Time

Item

Presenter

Notes

Time

Item

Presenter

Notes

5M

Action Item Follow Up

@Edward Zarecor


50M

Review xAPI mappings

@Edward Zarecor

  • Some discussion points have emerged from the review of this pr

  • I’ve attempted to incorporate the best of

  • Critical design questions:

    • Should we attempt to keep the same amount of information in xAPI documents that is currently available in the tracking logs.

    • What’s the right design for optional inclusion of data, for example, question data could only be included for randomized problems?

  • Discussion

    • @Former user (Deleted) very important to use existing extensions and when necessary we should propose a reusable extensions for the entire community to use.

    • @Julien Maupetit why is the approach to keep all the data that is currently in the events? Typically we should resolve the details via links.

    • @Zia Fazal What is the downside to adding an extension for new data like?

    • @Julien Maupetit We should favor lean first and only add data that does a job.

    • @Dave Ormsbee (Axim) points out that randomization of questions may require that provide the literal questions presented to learners because it would not be practical to reconstruct later – maybe not even possible.

    • @Julien Maupetit Can we make fields optional to drop data that isn’t required.

      • Event truncation and HTML in events is an ongoing problem.

    • @Andy Shultz (Deactivated) Debugging the stream

      • Having encoded fields make it hard for humans.

      • However, having full access to the data is useful

      • Even if we avoid putting the question, we will still need to record long free-text answers.

    • @Edward Zarecor How do other systems like Moodle handle, say, configurability of xAPI events

      • Moodle doesn’t handle this natively, but via plugins

      • What’s included if specified via profile

    • @Zia Fazal Can you we agree about an xAPI profile that is widely used and choose it.

    • Profiles and extensions are community driven, there isn’t a clear way to determining that one is preferable today. Adoption would be a determining factor, but it’s hard to measure currently

    • @Former user (Deleted) why not define our own Open edX specific profile rather than mapping?

      • A lot of vocabulary is missing on ADLNet

      • Would it be beneficial for the Open edX community to contribute to the xAPI community.

      • Would it be a plus for us to define a profile

      • Video profile for example, specifies one specific activity type more specifically.

    • @Julien Maupetit if we want to push forward with customized profiles we would need to have a real working sessions.

    • @Jill Vogel worries that we cannot specify in advance what everyone needs and we will end up not getting anything done.

    • @Julien Maupetit we shouldn’t need to specify everything in advance, but collaborating with the GaiaX group would prevent unnecessary divergence and duplicated effort.

    • @Jill Vogel We can have multiple versions, and add features as we go, but we should have use cases that drive what is included, not simply do this a a reverse engineering exercise.

    • @Julien Maupetit perhaps 1 Open edX event would be converted into multiple xAPI event, for example.

    • @Edward Zarecor should we start by gathering use cases from hint users to see if this feature is valuable enough now to care about?

    • @Dave Ormsbee (Axim) Looks like choice hints legacy is in part from LonCAPA

  • Why xAPI in Fun’s case

    • Additional flexibility of Caliper

    • Truly open community, no need to pay or join to have stake

    • No ability to define new verbs

10M

Call for interest in a Data Ethics subcommittee

@Edward Zarecor

  • Would someone be interested in leading the effort?

  • What, broadly, would it’s goals be?

 Action items

@Edward Zarecor Schedule a time for us to brainstorm architectural options for supporting per learner block level progress. [Carry over]
@Edward Zarecor Will reach out to MIT to understand the use of hinting data in research.

 Decisions