Meeting, 14 March 2011

Meeting, 14 March 2011

When

  • Monday 14/3/2011, 10:00 - 16:00

Where

  • DEIS Cesena, via Venezia 52, Meeting Room, 2nd Floor

Who

  1. Gabriella Castelli (UNIMORE)
  2. Marco Mamei (UNIMORE)
  3. Elena Nardini (UNIBO)
  4. Mirko Viroli (UNIBO)
  5. Marco Santarelli (UNIBO, master student)
  6. Andrea Omicini (UNIBO, only in the afternoon)
  7. Sara Montagna (UNIBO, only in the afternoon)

What

This meeting was aimed at sharing some ideas on a first prototype implementation of an LSA-space, based on the model of eco-laws described in a recent technical report TR.WP1.2011.2.pdf

— Elena Nardini's presentation

Elena recapped the main technical results of the TR.WP1.2011.2.pdf. This was useful to deepen the proposed model and start some discussion about the following issues:

  • The current structure of LSAs, featuring explicit ID and type, could be weakened by considering properties id and type
  • The issue of access restriction of LSAs is indeed an important one, but we feel we should stay orthogonal to it until the desired expressiveness of eco-laws has been achieved
  • There is an issue of garbage collecting LSAs no longer used, or references to LSAs no longer existing. 
  • Debugging applications could be tricky; solutions include: inspecting events inside an LSA-space, letting the infrastructure support agent-LSA interactions by proper scheletons
  • Eco-laws in the TR.WP1.2011.2.pdf are rather application-specific: it seems however this will be the scope of SAPERE in year 1
  • The role of ontologies has been discussed a bit, noting that it can be useful as a documentation of what agents can/should do in a SAPERE infrastructure, and that standard ontologies alone might be insufficient to realise certain tag-based matching scenarios

— Marco Santarelli's presentation

Marco Santarelli presented the work on his Master Thesis, that is, architecture, design and implementation of a prototype LSA-space. The following discussions on open issues followed:

  • There will be need (and there is indeed room) for studying optimisations to the reaction manager (this happens both for managing the event queue and for selecting candidate LSAs)
  • The architecture is open to an implementation of LSAs in terms of RDF (as proposed by STA)
  • Diffusion of LSAs could in the future be transmitted to the same port of standard operations

Marco Santarelli concluded presenting a demo based on the case study showed in TR.WP1.2011.2.pdf.

— Code Browsing

In the remaining time, we browsed the code and discussed several issues concerning the OO design of the LSA-space.

— Follow-up

The next step of this activity will be to study a methodology for sharing the code of this prototype and manage its evolution by any partner of the project.