HyperAgents / hmas

An ontology to describe Hypermedia Multi-Agent Systems, interactions, and organizations.
https://purl.org/hmas/
1 stars 0 forks source link

Can a hmas:HypermediaMASPlatform be hosted ? #122

Closed DrLeturc closed 1 year ago

DrLeturc commented 1 year ago

@andreiciortea wanted to reopen an issue to clarify about "Can a MASPlatform be hosted?"

There are many previous issues that have been closed related to :Hostable : https://github.com/HyperAgents/ns.hyperagents.org/issues?q=is%3Aissue+is%3Aclosed+Host

Even if the issues https://github.com/HyperAgents/ns.hyperagents.org/issues/8 https://github.com/HyperAgents/ns.hyperagents.org/issues/23 talk about "what can be hosted/nothosted", it seems the answer has not been provided .

Can we clarify our position on this ?

andreiciortea commented 1 year ago

@DrLeturc thanks for opening this issue!

I maintain the view from Issue #6 quoted also by @maximelefrancois86 in Issue #8 but stronger on the point that I don't see the usefulness of having an hmas:HypermediaMASPlatform hosting another hmas:HypermediaMASPlatform.

Examples of platforms that would fit the definition of hmas:HypermediaMASPlatform include Facebook, Twitter, Reddit, Yggdrasil, JaCaMo-REST, or Jade-hypermedia: all these platforms provide various features that support Hypermedia MAS, but I don't see how one could host another. Instead, they host agents (as running processes), artifacts, information resources, etc.

What introduces this problem in the core ontology is that hmas:HypermediaMASPlatform was defined as a subclass of hmas:Artifact. I don't agree with this definition either: there could be an hmas:Artifact that encapsulates one or more services provided by a platform but platforms such as the above-mentioned examples are not hmas:Artifacts (as the concept is currently defined in the core ontology).

The same applies to hmas:Workspace: we could represent a workspace artifact that agents can use to interact with the workspace (e.g., to create other artifacts within the workspace) but the workspace itself is not an hmas:Artifact.

andreiciortea commented 1 year ago

On the last two points, hmas:HypermediaMASPlatform and hmas:Workspace being subclasses of hmas:Artifact: this relates to Issue #32 — thanks @danaivach for surfacing that issue!

I think there is some confusion around these concepts and now that confusion bottoms out to the definition of hmas:isHostedOn. We need to revisit these points.

Going back to the discussion in Issue #32 and Issue #13: I agree still that only environmental resources should expose signifiers, but this does not exclude abstract concepts such as platforms and workspaces being reified through environmental resources (potentially artifacts) that expose signifiers.

andreiciortea commented 1 year ago

In the heartbeat meeting on Jan 25, we agreed to solve this issue by removing the :HypermediaMASPlatform and :Workspace as subclasses of :Artifact. I will open a PR with these changes.

DrLeturc commented 1 year ago

Need to be discussed during the plenary (15min)

DrLeturc commented 1 year ago

=> We accepted the corresponding PR.

To Do :