Progol source code and examples (16.3.09)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Contained in this directory is the source code, examples and man page
for the Inductive Logic Programming system Progol (Version 4.4),
together with a new tutorial on Progol4.4 by S. Muggleton and J. Firth
(this was updated and extended in July 1999). The tutorial is a
chapter which is to appear in the
book on ILP&KDD to be edited by Saso Dzeroski and Nada Lavrac.

Progol is freely available for academic research and teaching. Progol is
also available under license for commercial research. To apply for such a
license, please write to s.muggleton@imperial.ac.uk.

Progol4.4 extends Progol4.2 in the following ways.

1. Use of constraints for parameter estimation. This is an adaptation
	of the `lazy-evaluation' mechanism, first proposed by
	Srinivasan and Camacho. For instance, suppose we want to find
	upper bounds for the predicate lessthan/2. First,
	use declarations such as the following.

  :- modeb(1,lessthan(+float,#float))?
  :- constraint(lessthan/2)?

  The constraint declaration says that any lessthan/2 atom in the
  most-specific clause should have a Skolem constant in place of the upper
  bound constant (#float above). In the search, during any refinement which
  introduces a constraint atom, the flag solving is turned on, and the
  user-defined predicate is given all substitutions from positive and
  negative examples as a list of lists of lists(takes the form [P,N] where
  P is from the positive examples and N from the negatives, and P,N are
  lists of lists, each list giving all substitutions
  related to a particular example), and returns an appropriate
  substitution for the constant. This constant is used in place of the Skolem
  constant in subsequent testing of the refined clause and its
  refinements. Thus definitions for constraint predicates have to
  have at least two clauses, having the guards `solving' and
  `not(solving)' to define respectively the procedure for computing the
  parameter and the normal application of the predicate.

  Although originally not intended for such by Srinivasan and
  Camacho, the constraint solving mechanism can be applied to
  user-defined predicate invention. This is demonstrated in the release 4.4
  example files.
2. permute/1 can be used for randomly permuting the clauses in
  a file. This is useful for carrying out N-fold cross-validation.
3. The source directory contains qsample.c which can be used for
  randomly sampling training and test sets. Comments in qsample.c
  explain how to use the program.
4. The Clocksin and Mellish Prolog built-in predicate assertz/1 has been added.
5. The new built-in predicate aleave/1 carried out analogical prediction
	as described in the paper "Analogical Prediction" by Stephen Muggleton
	and Michael Bain (submitted to ILP'99).
6. A sample Unix script which runs on the animals.pl training set has
	been placed in examples4.2 to allow you to see whether you get
	the same output. The relevant files in that directory are as follows.

	Runanimals - Unix shell script to run progol on animals file
	animals.pl - training set (20 examples sampled with replacement)
	anim_test.pl - test set (151 examples sampled indepently from training
	set)
	animals.out - output from Runanimals run on a Linux Redhat Intel
				pentium PC.


Progol4.1 is fully described in the paper ``Inverse entailment and Progol''
by Stephen Muggleton which appears in the New Generation Computing
Journal, Vol. 13, pp. 245-286. This paper is available
as gzipped Postscript by anonymous ftp from ftp.cs.york.ac.uk in
file pub/ML_GROUP/Papers/InvEnt.ps.gz.

The mechanisms used for learning from positive examples are described
in the paper ``Learning from positive data'', which appeared in the
Proceedings of the ILP96 workshop, Springer 1997 and the paper
``Stochastic logic programming'' which appeared in ``Advances in
Inductive Logic Programming'' ed. L. De Raedt, Ohmsha/IOS Press, 1996.
These papers can both be obtained as Postscript by anonymous ftp from
ftp.cs.york.ac.uk in files pub/ML_GROUP/Papers/poslearn1.ps
and pub/ML_GROUP/Papers/slp.ps.

To get hold of Progol4.4 use ftp to copy the following
files to your directory.

	progol4_4.tar.gz
	expand.sh

Now do the following shell command:

$ expand.sh

This should produce the subdirectories, `examples4.2',  `examples4.4',
`man' and `source'. progol and qsample will be automatically compiled in
subdirectory `source'.

Include progol's source directory in your .login file path entry to allow
you to use progol as a command.  Remember to source your .login file before
continuing.

Try running Progol on all of the examples and see what happens.



*****************************************************************************
Professor Stephen H Muggleton            Tel: +44 20 7594 8307 (direct)
Computational Bioinformatics Laboratory  Bridget Gundry: +44 20 7594 1245
Department of Computing                  Fax: +44 20 7581 8024
Imperial College                         Email: shm@doc.ic.ac.uk
180 Queens Gate                          URL: http://www.doc.ic.ac.uk/~shm/
London SW7 2BZ
United Kingdom
*****************************************************************************