Working Group meeting

Date: 11/11/2021
Participants: Cecile Guasch, Giorgia Lodi, Hilde Kjølset, Natalie Muric, Giovanni Paolo Sellitto
Model editor: Eugen Costezki
Note editor: Andreea Pasăre

Agenda:

  • Merge into master is ready (backlog corrections of definitions: classes, attributes & relations)

  • Go through the errors discovered by the constraints checker

    • Missing target roles

    • Missing multiplicity

  • Present approaches to roles

  • What shall be used as context?

    • Context / Situation / Frame / Event / Process / Organisation

    • Most likely the context nature differ by “groups of roles”

    • Also, the context possibly differs by role participation (e.g. Buyer always participates) or role assignation (e.g. information provider is not participating but is assigned to offer information at some point in the future)

    • Also, the context differs by the role types, for example, organisational roles e.g. group leader, are different from participatory roles.

  • Shall we capture the situation frames or not?

    • Set of participants vs situation of a single agent

    • Distinction between actual(factual & time bound) & future(obligation, possibility, necessity)?

    • Frame of the event vs event outcome

      • E.g. Award Decision vs Awarding Action

    • Example: Awarding

      • the party that is performing the awarding: “the awarder” / “buyer”

      • the party that is awarded: “the awardee” / “winner”

      • the lot being awarded: “the object”

    • Contract signing

      • the signee on the side of the buyer: “…​” / buyer

      • the signee on the side of the winner: “…​” / winner

      • the contract being signed: “the object”

(Future agenda)

  • Discuss the principles on specifying the cardinality constraints (in the model and in AP) == Discussion

Presenting a summary of the last session’s discussion on roles and the various features that we previously adopted for roles.

In the literature, there are three approaches to modelling roles:

  1. Roles as named places - the most economical and the most straightforward

wfHTwcjscE1OwAAAABJRU5ErkJggg==

One limitation of this approach is that it is difficult to represent a group of lots.

  1. Roles as specializations and/or generalizations

AZ6O7rpg7hr4AAAAAElFTkSuQmCC
  1. Roles as n-ary relations

eZPSUYnmT4sAAAAASUVORK5CYII=

Discussing the CPSV-AP model.

We can organise the roles and subroles as classes and whenever we need to instantiate it we can always instantiate that class, so we don’t point to the enumeration.

We shouldn’t create subclasses for all the subroles. But do we really need the classes for roles, as well?

Is it useful to reify each role?
Roles may have states that’s why we should be able to capture the additional properties they will inherit.

Subroles as adjunct instances example:

1mKxRMwAAAABJRU5ErkJggg==

The EU Vocabulary Subroles includes group lead. How can we manage that?

Yqon3lL+VuuD9oAgiDItsEcyaoQGp6MOLKBqJSNaCtsgFwM52+2GqbzWZdtawPlBG0AQRAEKS8v1QZKB22AQRtAEARBys0mbQCEQGC5sdEnR46CDWypChD+H0tbq9wAJxgLAAAAAElFTkSuQmCC

We could bring the subroles into a controlled list and we will have a consistent approach to modelling Roles and Subroles.

Decisions:

  • Continue with the two parallel hierarchies.

  • Look at the current model and provide a solution of the main roles.

  • Test the solution with SPARQL queries.

  • Merge into master is ready (backlog corrections of definitions: classes, attributes & relations)