########################################################################

Support Vector Ordinal Regression with Explicit Constraints v1.0
Chu Wei (C)Copyright 2002-2004 at Gatsby Unit.

########################################################################
SOURCE CODE: http://www.gatsby.ucl.ac.uk/~chuwei/code/svorex.tar
########################################################################

1. Installation
	a) download svorex.tar to your machine (assume Linux)
	b) tar -xvf svorex.tar
	c) cd svorex
	d) gcc -o svorex *.c -lm -O2 -Wall
	
2. Data Format
	a) data should be stored as plain text files.
	b) each line contains one sample.
	c) elements are separated by a space.
	c) for each line, the last element is the target, integers from 1
	 to rank R; other elements are of the input vector.
	d) training data saved in a text file, say "mytask_train.0".
	e) the data for test with no targers saved in "mytask_test.0".
	f) the test targets are saved in "mytask_targets.0" if available.
	g) "mytask" could be any string followed by "_train" and "_test".
	
3. Input and Output
	a) prepare your data files
		"mytask_train.0"
		"mytask_test.0"
		"mytask_targets.0" (optional)
	b) run the program
		./svorex mytask_train.0
	c) 5-fold cross validation will be done for model selection.
	d) output files
		1) "mytask_cguess.0" to save predictive ordinal scales
		 for test data, integers from 1 to R.
		2) "mytask_cguess.0.svm.conf" to save the predictive
		 function values of test data.
		3) "mytask_train.0.svm.conf" to save the function values
		 of training data at the optimal solution.
		4) "mytask_train.0.validation" to save the validation
		 results of the function values for training data.
		5) "mytask_train.0.svm.alpha" to save the optimal
		 alphas and thresholds.
		6) "validation_explicit.log" to save the model we selected.
		
	e) if you have a BATCH of tasks, prepare your data files as
	 "mytask_train.0" "mytask_test.0" "mytask_targets.0"(optional),
	 "mytask_train.1" "mytask_test.1" "mytask_targets.1"(optional),
	 "mytask_train.2" "mytask_test.2" "mytask_targets.2"(optional),
	 ... ...
	 the program 'svorex' will make them done one by one.
	 the results of these files are saved in "ordinal_explicit.log".
	
4. Options
	a) for help information
		./svorex
	b) specify K-fold cross validation, say 10-fold
		./svorex -F 10 mytask_train.0
	c) specify the number of Rehearsal that could reduce the
	 randomness in fold generation, say 20 times
		./svorex -R 20 mytask_train.0
	d) specify the grid for searching in the space of model parameter
	 say C starting from 0.001 ending at 1000 with step 10^1
		./svorex -Cc 0.001 -Ce 1000 -Cs 1 mytask_train.0 
	 similar for Kappa (for Gaussian kernel only)
		./svorex -Kc 0.001 -Ke 1000 -Ks 1 mytask_train.0
	e) specify the times of 'zoom in' in grid searching, say 3 times
		./svorex -Z 3 mytask_train.0
	f) specify to use linear kernel 
		./svorex -P 1 mytask_train.0
	g) specify to use polynomial kernel with order 3
		./svorex -P 3 mytask_train.0
	h) normalize input features to normal distribution
		./svorex -i mytask_train.0
	i) seed the random number generator for reproducibility
		./svorex -S 2004 mytask_train.0

5. Reference
	"Support Vector Ordinal Regression",
	Chu, W and S. S. Keerthi, Technical Report, Yahoo! Research Labs 2005
	http://www.gatsby.ucl.ac.uk/~chuwei/paper/svor.pdf

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~