Overview of interactions

Q1. John McCarthy:

It seems to me that if you used formalized contexts, then you would be able to isolate different sets of propositions and say something about them in logic. In that way you wouldn't have to invent a new logic for this purpose.

A1. Renata Wasserman:

The difference is not all that great.

C1-1. Pat Hayes:

Yes, but if you reify formulae and quantify over them, then it is a new logic. (Agreement from McCarthy).

Q2. John McCarthy:

A compartment could be a context, couldn't it?

A2. Renata Wasserman:

We preferred to use the word "compartment" over "context".

Q3. Michail Soutchanski:

(Question not recorded)

A3. Renata Wasserman:

Another thing that we'd like to do is to facilitate retrieving the compartment. Predefined dependency relations between previous beliefs may be instrumental for that purpose.

Q4. Tom Costello:

(Beginning of question not recorded). ... I understand your logic treats conjunction differently from having the two conjuncts separately. Isn't this very strange?

A4. Renata Wasserman:

That depends on what you use for retrieving the relevance.

Q5. Bob Kowalski:

I was going to propose using metalogic, but then the problem is that you are naming sentences rather than propositions. A possible way out may be to use metalogic, but in such a way that it operates on a canonical form for the sentences, for example, clausal form.

A5. Renata Wasserman:

I don't think that we (people) represent sentences in canonical form.

C5-1. Pat Hayes:

But that's not incompatible.

C5-2. Graham White:

(Referring to the discussions on both questions 4 and 5) But there are situations where different representations are treated differently.

C5-3. Pat Hayes:

(To Graham White) Possibly, but that opens a can of worms. Imitating nonlogical behavior will carry too far.

Anyway, these logics have been studied argumentation theoretically, so there is plenty to choose from. If you don't like one semantics, then go back and pick another one.

C5-4. Erik Sandewall:

(To Hayes) That's dangerous advise. Changing semantics is like throwing a big switch on your whole system, with all its ways of representing different aspects of the application at hand. Unless, of course, you are able to interpret different parts of the system with different semantics - but that's a big can of worms as well.

C5-5. Tom Costello:

This all reminds me of the fights between programming languages, and the experience is that the worst language always wins.

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.