Temporal Usage Control Policy (UCP) execution for Solid Resources

[woutslabbinck]

Pitch

For the Sytadel project, we want to share the positional data of a vessel when consent is given by the skipper. The skipper is the captain of the ship and lives there, so he/she does not want to always share his positional data with anybody.

Therefore, a first Proof of Concept (PoC) to have temporal access control is preferred. Since the project is related to the International Data Space Association (IDSA) and there they have already defined some Usage Control Policies (UCPs) such as the Duration-restricted Data Usage Policy, it would be preferred if the implementation uses such a policy as a basis.

Desired solution

A demo where there is a service that can execute a Duration Usage Policy. Preferably accompanied with a screencast, so that it is easy to comprehend how it works.

The solution should follow the Koreografeye architecture. Which means that it needs both N3 rules and accompanying Plugins. As a result, the execution of this policies can be provided using a rule-based web agent such as the Solid Agent.

Acceptance criteria

• an N3 rule that has a duration UCP as premise and as inference a policy
• an actor/service that can do CRON jobs in combination with koreografeye
• a Koreografeye plugin that can execute a given policy
• a screencast showing a duration UCP in action. Furthermore, an explanation during/about the screencast must be provided.
• the Solid Resource has the correct access control as defined by the policy.

Pointers

• https://www.imec-int.com/en/sytadel
• https://internationaldataspaces.org/
• https://international-data-spaces-association.github.io/DataspaceConnector/Documentation/v6/UsageControl
• https://international-data-spaces-association.github.io/DataspaceConnector/Documentation/v6/UsageControl#duration-usage-2
• https://github.com/eyereasoner/Koreografeye
• https://github.com/woutslabbinck/Solid-Agent
• https://github.com/eyereasoner/Koreografeye-Plugin

[woutslabbinck]

An interesting question that was posed during an offline discussion:

What happens if data was used during the training of an Machine Learning Model, where due to the temporal access we do not have access to the raw data anymore?

[pheyvaer]

@woutslabbinck Looks great! One small thing about the screencast: the explanation should not be optional. A screencast of someone typing and clicking without any explanation is in most cases useless. Note that the explanation doesn't have to be audio. You can also use subtitles for example.

[argahsuknesib]

This is great. It is useful for the solid stream aggregator as well.

[woutslabbinck]

I have finished the demo, which is elaborated in following README on the cron-plugin branch in the Solid Agent code base.

This README.md contains the following information:

• A brief overview of the use case
• How to run the demo
• A screencast + explanation of what is happening
• Lower level explanation of the internal working of the Solid Agent in this configuration
• Information about the assumptions for this demo/implementation and the current limitations

If more information is required to finish this challenge, let me know!

[pheyvaer]

You find the report for this challenge here.