[CostAT] Panel sumarize

Panel about trust, argumentation and semantics. Francesca XXX, as moderator begins with 6 questions related with these 3 topics.

Jordi Sabater: Talking just about arguing about reputation. But this is not a panel, but a paper :-( I thought ‘panelists’ were going to give their own vision of the problem.

Eugenio Oliveira: He talks about trust. Very general things.

Carles Sierra: He asks for the 6 original question and talk about them and his own concept of trust, linked with argumentation and semantics all the time. Well done… this is interesting because he’s not trying to teach us, but to .

Piero Bonneti: About the intgegration of trust and negotiation models. How to do it nad for what?. Open issues, again involving us to participate because we can contribute to all these things. He’s focusing on large scale systems. I like this, because the problems that arise and the need of distributed solutions dealing with local information. But, at the end, he is going off the point.

Second panel after have lunch, is about norms, organizations and (again) semantics (it seems to be a very important thing, after all :-) Panelists are Olivier Boissier, Marco Colombetti, John-Jules Meyer and Axel Polleres. Better than this morning, Dealing with the questions one by one, assistants have the floor. But after the second question is getting too slow and a bit boring.

More participative than CAEPIA session. A lot to learn or just to show mutual interest and ‘scientific responsibility’?

Blogged with the Flock Browser

Tags: , , ,

[CostAT] WG1 meeting. Outreach and community building

A couple of presentations from yesterday. Michal S. begins talking about the use of agents to deal with grid resources. Things in common with semantic service. Actually, grid resources can be considered as services. Some points

  • teams (at least a very first core) are created off-line
  • just for local grids? It’s difficult to apply that to global grid (the Internet)
  • centralizzed CIC (matchmaker? useless
  • computational description of machines. But users want their work done and they don’t describe it in terms of CPU model and frequency.
  • trust and cold start problem
  • contract fulfillment
  • how can you check that SLA are met?
  • and real-time considerations too

Adina Magda continues introducing her group AI-MAS lab, in Bucharest Univ.

To do: prepare a document resuming two visions:

  • how SW can help to AT
  • how AT can help to SW

and that focused on the 3 topics that we’ve treated. I’m going to participate in the services part with Alberto and Adina.

Blogged with the Flock Browser

Tags: , , , ,

[CostAT] WG1 meeting. Presentation of STSMs

(See the schedule and additional information).

After review the tings done this year, some open issues:

  • more coordination, web site, mailing list and activities (besides STSMs and IE)
  • more networking with other WG
  • more networking with other communities, especially SW and social web

Today, we’re going to talk about

  • authority reasoning (Axel Polleres and Piero  Bonetti). Problem: dealing with billions of triples is slow, even with polynomial inference procedures, so the solution is deal with incompleteness, and authoritativeness is one of them (related with trust).
  • object consolidation (Antoine Zimmermann and George Vouros). Trying to complete information through consolidation/disambiguation.

Interesting things for me:

(Kifer & Subrahmanian, JLP, 1992) Mapping rules with confidence calues (annotated programs). Definition of a total order in \(CV=D_1 \times \ldots \times D_z \). Where for each domain \(D_i \) there is defined a total order relationship \(\leq_i \)which  leads to a order relationship \(\preceq\) in CV (total at the moment). I’d like to use this to formalize ordering in the vectors for consensus networks.

Instance matching. Given two ontologies \((S_1, A_1, I_1)\) and \((S_2, A_2, I_2)\), where \(S_i\) are signatures, \(A_i\) are assertions and \(I_i\) are instances, try to determine when two elements matches. This can be very useful to semantic service matching

Alberto Fernández and Michail Salampasis finishes the session today with an overview of their respective work, on semantice service discovery (alignment problem) and semantics for web annotation for blind people.

Blogged with the Flock Browser

Tags: , , , ,