[Prev][Next][Index][Thread]

Minutes June 3rd



Regarding the naming of subexpressions I'd like to point out two slightly
differing kinds of subexpression/expression naming relevant to producing
speech.

In AsTeR, you could name a cross-referencable object such as an equation with
meaningful names e.g. Fibonacci's  formula instead of equation 1.13
--this made listening to cross-referenced material easier.
See http://www.cs.cornell.edu/Info/People/raman/current/phd-thesis/html/node11.html#SECTION00162000000000000000

AsTeR also performed the other kind of sub-expression naming which I think got
discussed yesterday during the meeting.
I refered to this in AsTeR as "variable substitution".
--"folding" to use Dave's terminology--
AsTeR also implemented a fairly straight-forward naming algorithm for giving
meaningful names to sub-expressions e.g. "lower constraint 1" and denominator
3" etc.

The problem of recognizing that two subexpressions are identical is actually
very hard if you're working with a written/typeset document --AsTeR did not
attempt to do this. However, even without the intelligence to recognize two
semantically equivalent instances of a  subexpression, "variable substitution"
or "folding" as Dave refers to it in his minutes is *really* useful for speech
and I suspect it would be very useful for small visual displays.

In fact after I implemented the variable substitution  in AsTeR Rich Zippel at
Cornell pointed out that symbolic math systems in the early 70's had to
implement something analogous to display complex expressions since the
displays available at the time were not as graphically rich as what we have
today.

Here are the relevant sections from the thesis:

http://www.cs.cornell.edu/Info/People/raman/current/phd-thesis/html/node90.html#SECTION00467000000000000000
and
http://www.cs.cornell.edu/Info/People/raman/current/phd-thesis/html/node83.html#SECTION00460000000000000000

-- 
Best Regards,
--raman

      Adobe Systems                 Tel: 1 (415) 962 3945   (B-1 115)
      Advanced Technology Group     Fax: 1 (415) 962 6063 
      (E 1-160) 1585 Charleston Road       Email: raman@adobe.com 
      Mountain View, CA 94039 -7900  raman@cs.cornell.edu
      http://www-atg/People/Raman.html (Internal To Adobe)
      http://www.cs.cornell.edu/Info/People/raman/raman.html  (Cornell)
Disclaimer: The opinions expressed are my own and in no way should be taken
            as representative of my employer, Adobe Systems Inc.
____________________________________________________________________________


References: