Re: Clean Object Class Design -- What is it?

"Marvin mMnsky" <minsky_at_media.mit.edu> wrote in message news:f04e2625.0107231742.5485bbb_at_posting.google.com...

> Evidently they just scanned the old typewritten memo and postscripted
> it. Gah! Well I just OCR'd a published version and html'd it. So
> now it has OCR errors and some MSWORD "Save as HTML errors, too-but at
> least it is readable. It's at
>
> www.media.mit.edu/people/minsky/papers/Frames/frames.html

Not bad. 3 hours, 39 minutes. The MIT AI labs must be quite well-equipped (and well-staffed at almost 10 pm on a Monday night) ;-)

> >You likely will find no joy in the content either, as it is not a
> rigorous mathematical paper. If you are looking for a formal model of
> frame systems perhaps someone else can provide a link.
>
> Well, that is a priceless remark-seeing as many ideas in theat paper
> have never been formalized at all, e.g., the basic idea of
> 'frame-systems' (renamed 'frame-arrays" in Society of Mind). As a
> mathematician, I like to say that "Mathematics should be on tap, not
> on top"-because any particular formalization is likely to miss other
> important points. So most people think of frames as nothing more
> than property-lists, and miss the idea of introducing inheritance by
> re-assigning the values of default-assignments. Also, I think this
> may be the first paper that complains that conventional logical
> formalisms are too "monotonic."

At the risk of stepping way outside of my depth, after a quick google search found an interesting discussion on the PSU FOM (first-order maths) list. Here is a message from that thread:

http://www.math.psu.edu/simpson/fom/postings/0008/msg00048.html

The relational model is based on set theory. Language models like Smalltalk, Lisp, and the functional languages are based on lambda calculus. Set theory is not expressive enough to represent the semantics of second-order maths. Second-order types of expressions in the relational model could be important queries like "is this schema in 3NF?".

Now if you implemented the relational model on top of a Lisp or a Smalltalk back end, then you could formulate second-order queries using those language models. Also, instead of representing relational queries as Strings, they would be represented as lambda expressions. This would allow things like, for example, a query engine that could remember and recognize similarities among queries in order to provide more efficient and intelligent responses, much in the same way an information officer at a train station could answer similar queries about different train schedules during the course of a day (without going through the same mechanical lookup process each time). IOW, databases that learn.

If I am not totally off-base with the above, I think it provides a reasonable response to Bob's concerns about a lack of a formalism for OO models and how they can be useful in a relational context. As to the relationship between lambda calculus and OO, I'll pass the baton on that one.

--
Cheers,
Steve (steve at wart dot ca ICQ 50919689)