Chaired and scribed by Dave
- Dave Raggett
- Javier Rodriguez
- Cristina Cachon
- Nikolas Kaklanis
- Sebastian Feuerstack
- Davide Spano
- Fabio Paterno
- Jean Vanderdonckt
- Vivian Motti
- Francois Beuvens
- Pascal Beaujeant
- Jaroslav Pullman
- Gerrit Meixner
- Gaelle Calvary
Reminder: everyone to read through the editor's drafts and bring their comments to the face to face!
UCL: we have updated the gdoc for the AUI with the first version of the meta model. We are addressing the comments from Sebastian.
Jean: what is the most appropriate methodology tor each the goal, e.g. bottom up or top down when it comes
Fabio: so far it seems that you are following a top down approach.
Jean: we have been trying to take into account all of the partners input, but this takes a lot of time.
Fabio: we explain why we put in the task model document where we disagree with a comment
Jean: we need a consistent approach across subgroups
Fabio/Davide: we discussed a bit for the task model and for the AUI. We tried to keep the major concepts with an extensional approach, and minor concepts as intensional.
Jean: for which part, give us an example?
Davide: task categories as extensional and task types as intensional
Fabio: this seems reasonable
Jean: for the AUI if we take the category "select" I would have as many subclasses as I have ...
Jean: intension provides the most concise representation in the meta model
Jean: we have a choice between adding a new subclass or adding to an ennumeration
Fabio: at some point for the sake of brevity we would choose intensional. It depends on the level of granularity we are considering. It would be better to extend the main concepts.
Jean: we can model the same concepts in many different ways.
Fabio: the main concepts can be extended, and the low level concepts can be intensive
Dave: perhaps this is something we need to resolve at the face to face as it feels like it will be hard to reach a consensus on the phone?
Jean: you have as many subclasses as you have types, right?
Jean: for the metamodel in the google doc, there are 3 subclasses without any attributes
Fabio: no, these are associations, not subtypes
Jean: In AI there is a nice classification of user tasks, I might like to add these
Fabio: the diagram isn't ideal as a way to express extensibility
Jean: definition by extension
Fabio: we keep the categories and the attribute values can be extended.
Jean: google KADS task types for details, I don't remember exactly. For the operators it is different, right?
Jean: you want to modify the precondition in the choice?
Fabio: non-deterministic means a choice without a precondition
Davide: for a deterministic choice, if you define this through conditions, you don't need a deterministic choice, but you do need to represent the conditions
Jean: in the UK, ...
Fabio: boolean conditions for deterministic choices, without need for subclasses.
Further discussion about concurrent and parallel task models.
Jean: the main question we have is about the methodology, so that the standard has the greatest coverage?
Fabio: we want to avoid too many operators. We have a table that compares different notations, at least for the ones we have looked at so far.
Jean: for the line CTS, there are some that are not in the table
Fabio: which ones?
Jean: ... concurency is interleaved or parallel?
Fabio: so far we have left this in an abstract way so that it can be left to the context
Jean: if I want to choose between one or the other, how do we do that?
Fabio: if we think this is important, we can elaborate the distinctions
Jean: what process do we need to achieve a consensus?
Dave: we can do this at the face to face - to get people to present different approaches and to then see if we can reach a consensus on a synthesis or to pick the best approach, noting the reasons for doing so.
Jean: order independence is a particular case of concurrency?
Jean: concurrency is the most abstract class?
Fabio: there are two possible implementations of concurrency, but is it important to capture which at the level of task models
Jean: perhaps not at an early stage of the design work
Jean: to come back to the methods, you have used a bottom up approach for the task models, right?
Jean: we've used a top down approach for the AUI which is very time consuming. Should we switch to a bottom up approach?
Fabio: perhaps you should make a proposal, and we can see
Jean asks Dave for his views on how to proceed
Dave: we don't need to finalize everything right away, and first publication of the drafts is the immediate priority. We should aim for stable specifications towards the end of this year.
Jean asks for comment on the new version of the AUI in the google doc.
Davide: it seems to be in the right direction
No more comments on the AUI.
Dave any more to say on the Task model specification?
Fabio: no, we have already covered that, but please read through and bring your comments to Pisa.
Any other business
Jean: will you have a skype connection at the face to face?
Fabio: yes, we can arrange that