TITLE: The Rules of Modelling: Automatic Generation of Constraint Programs

PRESENTER: Alan M. Frisch, http://www-users.cs.york.ac.uk/~frisch/, frisch@cs.york.ac.uk

AFFILIATION:Department of Computer Science, University of York, http://www.cs.york.ac.uk/index.htm

DATE: Monday 2nd June 2008

TIME: 12:00:00

PLACE: CSE Seminar Room, Level 1, K17

ABSTRACT:

Many and diverse search problems have been solved with great success
using constraint programming. However, to employ constraint
programming technology to solve a problem, the problem first must be
characterised, or modelled, by a set of constraints that its solutions
must satisfy. Generating a correct model can be difficult; generating
one that is easier to solve than its alternatives is even more
difficult, often requiring considerable expertise. This so-called
"modelling bottleneck" has inhibited the wider use of constraint
programming technology.

This talk describes CONJURE, a rule-based system that automatically
generates constraint programs by refining an abstract problem
specification. Since the high-level specification language is
significantly closer than a constraint program to the way in which
problems are commonly conceived, the modelling bottleneck is
substantially reduced. A particular focus of this talk is showing why
the refinement rules must be recursive, why this is difficult to
achieve and how we ultimately solved this problem.

This talk assumes no background in constraint programming.

BIOGRAPHY OF SPEAKER:

Dr Alan M Frisch is a Reader in Intelligent Systems and Head of the
Artificial Intelligence Group in the Dept. of Computer Science at the
Univ. of York. He previously held faculty positions at the Univ. of
Illinois at Urbana-Champaign and at the Univ. of Sussex and has held
visiting research positions at Max Planck Institute for Informatics,
Univ. of Leeds, IBM T J Watson Research Center, AT&T Bell Laboratories
and Hokkaido Univ. For over 25 years he has been publishing
research on a wide range of topics in the area of constraint
programming, Boolean satisfiability, automated reasoning (including
the integration of constraint solvers in deductive systems), logic
programming, probabilistic inference, knowledge retrieval, and
inductive generalisation. His current work focuses on developing
a systematic and automatic account of the process
of modelling problems with constraints.

Dr Frisch is now on sabbatical working with the G12 Group at the
University of Melbourne.

Host:

Michael Maher

Seminar Convenor:

Van Hai Ho