home / SSP Talks

My SSP talks

• 28th November 2008: Learning from OpenKnowledge. A richer coordination framework based on LCC    [.pdf file]
  

  LCC has been used, in the course of the OpenKnowledge project, to model interactions of varying complexity.
  One of the difficulties I have often encountered, at least in the OpenKnowledge version of LCC, is the lack of
  an abstraction mechanism to partition the interaction into layers of different granularity. Roles are useful,
  but are often not enough: the whole interaction must be defined in detail in a single model.
  The lack of a proper layering mechanism also hinders the flexibility of systems like OpenKnowledge.
  In the real world it may be useful to define the higher levels of an interaction,
  and leave the detail for later, when the peers have started the interaction.
  This is hard to represent without an adeguate abstraction mechanism: the OpenKnowledge kernel offers a
  solution, but it's more of an implementation work-around than a clean model.
  Moreover, the relation between an interaction model and its execution by a group of peers is not well defined.
  In this talk I will try to address these limitations, presenting a proposal for extending LCC with the
  use of 'scenes' for layering interactions, and of 'tasks' for representing the
  commitment of peers to perform together a scene.
  

• 12th June 2007: An introduction to the OpenKnowledge kernel   [.pdf file]
  

  In this talk I will explain how the OpenKnowledge kernel is structured and how it works.
  The OpenKnowledge kernel is the implemented "playground" for testing and experimenting the ideas
  developed within the OpenKnowledge project. It is a peer-to-peer framework focused around the interaction models,
  that are specified as LCC protocols.

• 9th May 2006: Dynamic Ontology Mapping: a first prototype   [.pdf file]
  

  Within a MAS, individual agents interact in order to perform set of given tasks; for example buying
  a product, retrive information, and so on.
  Open MAS place weaker constraints on the agents that can take part in the interactions:
  often it is not possible to ensure that all agents share the same ontology.
  Mapping in advance all the possible combinations of ontologies may not be feasible.
  My proposal is to model the dialogues, as they take place: these models are used to increase the
  efficiency of standard ontology mapping methods.
  Mapping takes place when a new term appears in a received message, and the corresponding (or the closest)
  term in the agent's ontology is found. The model is used to predict the set of possible corresponding
  terms, reducing the number of computationally expensive comparisons between the received term and
  the terms in the agent's ontology.

• 8th November 2005: An Introduction to Uncertainty in Ontology Mapping (Part 2)   [.pdf file]
  

  In this second part, I will talk about handling vagueness through Fuzzy logic and fuzzy sets.
  I will also discuss Possibility theory and, briefly, argumentation logic.
  Finally, I will talk about how I could exploit these methods in my approach in Ontology Mapping,
  with the aim of identifying the most apt.

• 1st November 2005: An Introduction to Uncertainty in Ontology Mapping (Part 1)   [.pdf file]
  

  In this first part, I will give a description of what uncertainty is and I will
  present some quantitative methods for representing uncertainty. In particular,
  I will describe the probabilistic approach, Bayesian networks, the Certainty Factor, and Dempster-Shafer.

• 23rd August 2005: Exploiting Interaction Contexts for Dynamic Ontology Mapping   [.pdf file]
  Thesis Proposal
  

  This proposal addresses the problem of ontology heterogeneity in open multi-agent systems.
  In open MAS, it is often an impossible requirement that all agents share an ontology or
  that all the combination of mappings between different ontologies are foreseen.
  My proposal starts from the pragmatic assumption that understanding between agents is needed only to
  perform the interactions required for reaching some common goal. I propose that only small portions of ontologies,
  relevant to the context of the interactions are mapped.
  The mapping framework exploits the experience gained in previous interactions to quickly classify the
  dialogue into a context and then uses the context to filter out the unrelated parts of the ontologies.
  The mapping itself is modelled as decision-making under uncertainty.

• 1st February 2005: Dynamic Ontology Mapping in MAS   [.pdf file]
  

  In open Multi Agent Systems it is unlikely that agents developed by different communities share the same ontologies.
  This lack of homogeneity can make communication very difficult, as agents may fail to understand
  properly the messages they receive. Some form of agreement over the content of the message might be needed.
  In this talk I will propose some reference ideas that I hope will help me in my research
  in the field of mutual understanding between agents.
  I will also present a small example to clarify some of the concepts introduced.