Teaching ideas about grammar, parsing and generation,
Using Pop11's LIB Grammar package and extensions
(DRAFT: Liable to change)
Installed: 15 Nov 2013
Last updated: 17 Nov 2013
This web page is
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/grammar-sem.html
A partial index of discussion notes is in
http://www.cs.bham.ac.uk/research/projects/cogaff/misc/AREADME.html
Abstract for Possible IRLab Presentation
School of Computer Science
Monday 18th Nov 11-12am Room 245
FIRST DRAFT ABSTRACT -- TO BE REVISED AND EXTENDED
In this seminar I shall
-
give an introduction to a simple toolkit that can be used to add some natural
language interaction to a robot,
-
demonstrate a toy grammar for describing and interacting with a robot,
-
illustrate how extending a parser with an ill-formed substring table (as opposed to
the better known well-formed substring table mechanism) made a huge difference to
performance in parsing.
-
There's a general principle there that's widely applicable, regarding forms of
learning, i.e. the importance of learning what doesn't work and being able to
spot early steps towards failures. (One of the main reasons for doing lots and lots
of examples when learning mathematics, logic, programming, etc.)
BACKGROUND INFORMATION
There's an online video demo of an application of the grammar package included in
http://www.cs.bham.ac.uk/research/projects/poplog/figs/simagent/
Look at the two videos in Section 12,
demonstrating a very simplified version of
Terry Winograd's ground-breaking SHRDLU program, presented in his MIT PhD thesis over
four decades ago: http://hci.stanford.edu/~winograd/shrdlu/ .
For people who have never done any programming using grammars, either for sentence
input or for sentence output, or for both, it may worth looking at the online
video tutorials concerned with language, grammars, and chatbots here:
http://www.cs.bham.ac.uk/research/projects/poplog/cas-ai/video-tutorials.html
NB: these are all 'first draft' toy demonstrations that need to be redone. I
was still learning to use 'RecordMyDesktop', and the tutorials were unscripted.
The general philosophy of teaching programming behind these demonstrations, focusing
on "thinky programming" is outlined here.
The more elaborate, but still 'toy' experimental grammar for interacting with a robot
is presented here:
http://www.cs.bham.ac.uk/research/projects/poplog/examples/robolang/
Maintained by
Aaron Sloman
School of Computer Science
The University of Birmingham