This page is a historical archive. For the latest information please visit

Fausto Giunchiglia and Chiara Ghidini

Local Model Semantics, or Contextual Reasoning = Locality + Compatibility.

[mail to author]
[mail to moderator]
[debate procedure]

Overview of interactions

N:o Question Answer(s) Continued discussion
1 9.1  Tom Costello
9.1  Chiara Ghidini
9.1  Tom Costello
9.1  Pat Hayes
2 9.1  Erik Sandewall
9.1  Chiara Ghidini
3 9.1  Dov M. Gabbay
  9.1  Pat Hayes
9.1  Dov M. Gabbay
9.1  Pat Hayes
9.1  Dov M. Gabbay
9.1  Graham White
4 9.1  John McCarthy
5 9.1  Tom Costello
  9.1  Fiora Pirri
6 9.1  John McCarthy
  9.1  Fiora Pirri
9.1  John McCarthy
9.1  Pat Hayes
7 9.1  Vladimir Lifschitz
9.1  Chiara Ghidini

Q1. Tom Costello:

You define your semantics in such a way that each viewpoint has a separate interpretation. I would have imagined using a single interpretation to which all the viewpoints can be related. In your six-box example, there would have been one box with six compartments for explicating both views.

A1. Chiara Ghidini:

Usually we don't have the complete description of the world.

C1-1. Tom Costello:

But a semantics is not supposed to work like that. It need not identify a single model; it defines a structure, and a given set of axioms may have a very large set of models, reflecting our lack of knowledge.

C1-2. Pat Hayes:

I agree with Tom (Costello); the point is that there is only one reality.

Q2. Erik Sandewall:

Regarding question 1, I want to support Chiara's approach. If we were to follow Tom's (Costello) advise, we would have to define how the unique interpretation is projected into the different views. That is not necessarily any simpler than using the compatibility relations.

My question regards what is the syntactic counterpart of the compatibility relations. It would seem that a given set of bridge rules corresponds to certain compatibility relations, and vice versa. Is this correct, and if so, can you characterize the relationship between bridge rules and compatibility relations?

A2. Chiara Ghidini:

Yes, for limited cases of the bridge rules we can do that.

Q3. Dov M. Gabbay:

I also want to defend Chiara's approach. This corresponds to the fibering of systems (...)

C3-1. Pat Hayes:

(Interrupting Gabbay after a while): What in the world are you talking about? (...)

C3-2. Dov M. Gabbay:

The thing is that you and the speaker don't understand each other...

C3-3. Pat Hayes:

I don't understand you either. (...)

C3-4. Dov M. Gabbay:

The whole thing is about mixing two languages. You evaluate one until you reach an expression in the other, then you transfer to the interpretation of the other language (...)

C3-5. Graham White:

It would be interesting to know when it is possible to derive a global model from the compatibility relation.

Q4. John McCarthy:

In support of Chiara's approach, I also don't believe it's possible to assume a single representation of the world. Given my representation of Pat's beliefs and Dov's representation of Pat's beliefs, is there underlying that a true account of Pat's beliefs?

Q5. Tom Costello:

Comparing the two approaches of [c-fcs-98-114] (Pirri) and [c-fcs-98-75] (Ghidini), it seems that they are addressing similar topics but using opposite methods. Fiora (Pirri) decomposes a large thing into its parts, Chiara (Ghidini) takes local information and puts it together. The same words are being used for things that work in opposite directions.


c-fcs-98-114Gianni Amati and Fiora Pirri.
Contexts as relativized definitions: a formalization via fixed points. [postscript] [discussion]
Proc. Formalization of Commonsense Reasoning, 1998, pp. 114-125.
c-fcs-98-75Fausto Giunchiglia and Chiara Ghidini.
Local Model Semantics, or Contextual Reasoning = Locality + Compatibility. [postscript] [discussion]
Proc. Formalization of Commonsense Reasoning, 1998, pp. 75-86.

C5-1. Fiora Pirri:

The only difference is that Chiara can use two different languages; I just use one language.

Q6. John McCarthy:

Chiara started out by saying "a context is...", but her definition seemed much more narrow than what was in my original paper about contexts. This is a general complaint I have about this kind of work - it is rarely concerned with lifting a formula from one context to another. There's a lot of other things that need doing as well.

C6-1. Fiora Pirri:

I agree, but there is a problem with interpreting terms in contexts. A question to John again: how narrow is a context?

C6-2. John McCarthy:

If I make a definition of the form "A context is..." then I would give a definition that's more narrow than what I'd be prepared to live with.

C6-3. Pat Hayes:

At a recent workshop on context I collected definitions of what it is. They ranged from "a cultural viewpoint", via "a first-order theory" to "a point in spacetime". It seems to be difficult to get any agreement on that.

Q7. Vladimir Lifschitz:

It is interesting to compare this paper with John's two approximations (?). Here there are no approximations. An interesting alternative would be to consider two observers, both of which see everything but with observation errors. Do you envisage considering this case?

A7. Chiara Ghidini:

Yes, in federated databases there are examples of that.

This on-line debate page is part of a discussion at recent workshop; similar pages are set up for each of the workshop articles. The discussion is organized by the area Reasoning about Actions and Change within the Electronic Transactions on Artificial Intelligence (ETAI).

To contribute, please click [mail to moderator] above and send your question or comment as an E-mail message.