March 06, 2020: FLOE infrastructure

Present: Cindy, Colin, Jutta, Ned, Philip, Ted

Video Recording

Google Doc

Notes:

Not a one-size-fits-all across all resources but rather to match or find resources to match a particular learning goal. 

Exploration for students to discover their learning needs.

A way to reach out to vendors to deliver various forms of learning material.

Funders can have the data of learning gaps.

Difficulties with metadata:

  1. Manual metadata is crappy. Ppl don’t do it right
  2. Not extensive enough to cover learning goals


FLOE diagram: https://wiki.fluidproject.org/display/fluid/%28Floe%29+Diagrams?preview=/24945059/30048450/FLOE.jpg

There should be a more recent diagram

GOORU(http://gooru.org/about/): a search engine for OER

Colin asked about who: Informal self study groups for lifelong learning or closer to more formalized education (schools and classes).

Jutta: OSEP (Special Education Department in US) asked us to revisit the notion of teachers to gather content packages of a number of resources to meet the needs of their students. This would be a good use case. They are asked to work with some schools here that can provide formal requirements, testing and timelines. Issues with formal schools:

  1. Cannot break at what students learn and what they want to learn
  2. Where do learning resources come from, from federal repository or wild web? Or are they within the gated network


Ted: what data should we collect and present

Jutta: we don’t know what the whole set is. That’s why students need to constantly reflect and give feedback on it

Philip: Relating to this city work, there’s a pretty interesting discussion in this paper about what happens when citizens are empowered to critique their urban environment: http://www.klokmose.net/clemens/wp-content/uploads/2016/06/PDC2016-final.pdf
section 5.2

Other relevant IDRC projects to look at: Nexus, discovery tools, gpii, floe, myl3, social justice repair kit, CISL, toolkit for justice

Ted: AI and machine learning can be used in WeCount to recommend tools. The bias of AI in recommending is relatively fair and can be customized to particular users.

The goal of FLOE infrastructure:

  1. Explore how we can use AI to help with making learning more accessible
  2. Gaining insides and outsides of data science, its risks and problems
  3. Serve students and teachers


The starting point:

  1. Designers: Reach out to a learning group who have learning goals. Understand their learning requests and build use cases.
  2. Developers: discover functionalities from other projects that we can reuse or repurpose as well as functionalities that we need but haven’t developed.


An overview of FLOE infrastructure:

  1. Build based on FLOE diagram: https://wiki.fluidproject.org/display/fluid/%28Floe%29+Diagrams?preview=/24945059/30048450/FLOE.jpg
  2. Learners have their learning goals. These goals can be functional goals (learn it when you do) or notional goals (learning particular subjects such as art)
  3. Provide tools to help learners to explore and discover their learning needs, such as using glossaries, examples, pairing up with someone, or a late night learner. Learners should be able to configure what parameters they’d like to record or track.
  4. Matching engine: create a non-black box, transparent system that gives ppl control and choices to automate the decisions, which is important for ppl who are not represented by the standard data set. AI will be used in this engine to find learning material and tools to meet particular learning goals. The matching results for notional topics should wind down to presentation or non-subject oriented parts of it so learners can consume and understand faster.
  5. Learners provide feedback to reflect on what works well or doesn’t work well for them. The matching engine will improve results based on their feedback.


One possible outcome of this project could be:

  1. Pick a particular use case.
  2. Implement a data science tool around this scenario where we can explore risks and problems of the application of data science.