On Saturday, March 3, 2012 11:02:39 AM UTC-8, kcrisman wrote:
...
>
> Has anyone ever done a natural-language frontend attempt to Maxima or
> its predecessors? I would be surprised if someone hadn't, to be
> honest.
>
> I am unaware of any natural language front end to Macsyma or Maxima,
at least ones that became popular for users of Macsyma or Maxima.
As background ...
There is the thesis work of Daniel G. Bobrow on a program named STUDENT
that could solve some algebra word problems by a bunch of clever heuristics.
More recently there was a program that solved some physics word problems,
written by
Gordon Novak.
Some Googling should get you more references and citations to these programs
as well as current activity.
Basically the task that others in the past thought made sense was to take
a problem domain
that is naturally related to language and convert problems expressed in that
language to calculation: arithmetic or algebra or calculus.
Taking arithmetic or algebra or calculus and expressing it in natural
language strikes
me as pretty much useless ("please tell me the greatest common divisor of
three and four").
It appears to be both clumsy and backwards.
Compare that to "A train leaves Philadelphia toward New York at at the
same time as
a train leaves New York toward Philadelphia. ... etc. ". converted to a
system of
algebraic equations to solve.
Since Bobrow's work (circa 1963?) could easily be reproduced today as a
front end
to (say) Maxima, it could be demonstrated. I've suggested it as an
undergraduate
programming project in the past, but no takers. Maybe someone in the Sage
community would bite. I think it was written in a string-processing
language built on Lisp.
Novak's work was in Lisp too. But either could be done in python no doubt.
Looking at speech input work at Kingston University (UK) suggests that they
found somewhat the same problems as I found (earlier?) in my math speak
project, and probably did not get further. Hard to tell though.
I allowed x+z pronounced eks plus zee. they seem to show
x-ray plus zulu.
google for KU-talk
The government buy-in for speech is that it can be used by persons who
cannot
type, so you get money for disability-related accommodations, potentially.
Typing natural language like "greatest common divisor" is not an
accommodation
to anyone.
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
