Progol source code and examples (21.09.00)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Contained in this directory is the source code, examples and man page
for the Inductive Logic Programming system Progol (Version 4.5),
described in Stephen Muggleton's paper "Learning Stochastic Logic Programs"
in the AAAI2000 workshop on Learning Stochastic Relation Models
(paper is available from
ftp://ftp.cs.york.ac.uk/pub/ML_GROUP/Papers/slplearn.ps.gz).

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 stephen@cs.york.ac.uk.

Progol4.5 extends Progol4.4 in its incorporation of user-defined
evaluation functions. The following table shows the names of predicates
used to allow use of these evaluation functions.

Variable	Meaning			Built-in	User-defined
p		Positive coverage	pos_cover(P1)	user_pos_cover(P2)
n		Negative coverage	neg_cover(N1)	user_neg_cover(N2)
c		Hypothesis size		hyp_size(C1)	user_hyp_size(C2)
h		Hypothesis remainder	hyp_rem(H1)	user_hyp_rem(H2)

For instance, in order to implement a maximum likelihood search user_hyp_size
is set to a constant, eg by including the following in the Progol input file.

	user_hyp_size(1).

The first version of Progol, 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 Progol4.5 use ftp to copy the following
files to your directory.

	progol4_5.tar.gz
	expand.sh

Now do the following shell command:

$ expand.sh

This should produce the subdirectories, `examples4.2',  `examples4.4',
`examples4.5', `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.

==

	Stephen Muggleton,
	Department of Computer Science,
	University of York, Heslington, York, YO1 5DD.
	Tel: 01904 434750, Fax: 01904 432767,
	Email: stephen@cs.york.ac.uk.
	WWW: http://www.cs.york.ac.uk/~stephen