WSML Meeting 23-April-2004

"Meeting Protocoll"


Participants:

Dieter Fensel,  Michael Kifer, Uwe Keller, Holger Lausen, Francisco Jose Martin-Recuerda Moyano, Eyal Oren, Livia Predoiu, Axel Polleres, Dumitru Roman, Ioan Toma, (Miahesl Stollberg joining in the end)

Activity Sheet:

D5.1, D5.2:  new deadlines 17th of May,
D8: new deadline 3rd of May,
the rest of the activity sheet is ok.

Deliverables:

1) D3.2 (WSMO)

Discussion in the WSML meeting, since the modelling issues for D3.2 are also crucial for WSML.
Short briefing  for Miahael K. on the state of the modelling discussion. Michael and the authors of D3.2 agree to have a Telephone conference early next week.

2) D5.1, D5.2

Difference between D5.1 and 5.2: D5.2 is more on the theoretical proof obligations while 5.2 shall discuss tools and how to actually implement
these proof obligations. D5.2 is at the moment almost empty.

Details on the discussion on D5.1:
- General comment: currently only offline aspects, but no runtime discovery is discussed inth this deliverable.
- Dieter points to a reference of Ian Horrocks on matchmaking (cf. presentation by Eyal in the last WSML telephone conference).
- Comment: composition makes use of discovery, not the other way around. The general way to go shall be: discover- decompose - adapt - construct
  Dieter suggests to have a look at papaers on parametric design.
- Section 3.2: Proof obligations so far are not sufficient for dynamic composition as such, because you cannot just put together the conditions of the separate services. For the moment only one matching service which matches the goal completely shall be considered.
- Discussion on assumptions: What is the difference between proofing and checking?
- Section 4.1.1: Discussion on matching: A service is only usable if it always ends up in a state that implies the goal, the recution should provide what is missing to achieve such a match. In this context, there is a discussion on non-deterministic vs. deterministic services.
- Agreement: Simplify in the deliverable: ONE capability describes ONE service, an extension of this will be that one capability describes a set of services which is how it is fedined in WSMO
- Discussion in Section 4 somehow misleading: The discussion in this deliverable should not talk about the relation between matching between the capability and the actual service execution but only about goal-capability matching
- Structure of section 4 shall change
- What kinds of matching can be dane by LP, what not? Triple should also be mentioned here, however, all these points more focusing on implementation shall go to D5.2
- Uwe will include all comments and also try to incorporate the outcomes of the D3.2 discussion

Next Meetings:

Dieter will present WSMO/WSML in one of the next SWSI telephone conferences, Michael K. suggests June 3rd, 7:30 Ibk. time.
(Michael and Dieter will arrange this)

Due to travelling in the next few weeks, the next WSML telephone conference will be May 19th 4pm-7pm (Innsbruck time).

Presentations:

Eyal: "Representing Disjunction and Quantifiers in RDF"
Axel: "Formalizing Web Service Choreographies"