cssdiff(1) [suse man page]
cssdiff(1) CRM114 cssdiff(1) NAME cssdiff - generate a difference summary of two .css files SYNOPSIS cssdiff [cssfile 1] [cssfile 2] WARNING This man page is taken from an older CRM114 version. It is provided as a convenience to Debian users and may not be up-to-date. If you would like to update it, please send appropriate patches to the Debian bug tracking system. DESCRIPTION cssdiff is a special-purpose utility to measure the distance between the classes represented by cssfile1 and cssfile2. The summary result output tells how many features were in each of the .css files, how many features appeared in both (balanced overlap), how many features appeared only in one (or unbalanced overlaps), and how often the feature set of one .css file strictly dominated the feature set of another .css file. This set of metrics provides an intuitive way to determine the similarity (or dissimilarity) of two classes represented as .css files. When using the CRM114 spamfilter, it can be used to find out how easy it will be for CRM114 to differentiate spam from nonspam with your .css files. cssdiff prints a report like e.g. Sparse spectra file spam.css has 1048577 bins total Sparse spectra file nonspam.css has 1048577 bins total File 1 total features : 1605968 File 2 total features : 1045152 Similarities between files : 142039 Differences between files : 1279964 File 1 dominates file 2 : 1463929 File 2 dominates file 1 : 903113 Note that in this case there's a big difference between the two files; in this case there are about 10 times as many differences between the two files as there are similarities. OPTIONS There are no options to cssdiff. SHORTCOMINGS Note that cssdiff as of version 20040816 is NOT capable of dealing with the CRM114 Winnow classifier's floating-point .cow files. Worse, cssdiff is unaware of it's shortcomings, and will try anyway. The only recourse is to be aware of this issue and not use cssdiff on Winnow classifier floating point .cow format files. HOMEPAGE AND REPORTING BUGS http://crm114.sourceforge.net/ VERSION This manpage: $Id: cssdiff.azm,v 1.5 2004/08/19 09:06:44 vanbaal Exp $ This manpage describes cssdiff as shipped with crm114 version 20040816.BlameClockworkOrange. AUTHOR William S. Yerazunis. Manpage typesetting by Joost van Baal and Shalendra Chhabra COPYRIGHT Copyright (C) 2001, 2002, 2003, 2004 William S. Yerazunis. This is free software, copyrighted under the FSF's GPL. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the file COPYING for more details. SEE ALSO cssutil(1), cssmerge(1), crm(1), cssmerge(1) cssdiff 20040816.BlameClockworkOrange-auto.3 19 Aug 2004 cssdiff(1)
Check Out this Related Man Page
cssutil(1) CRM114 cssutil(1) NAME cssutil - utility to measure and manipulate CRM114 statistics files. SYNOPSIS cssutil [.css file] [OPTIONS] WARNING This man page is taken from an older CRM114 version. It is provided as a convenience to Debian users and may not be up-to-date. If you would like to update it, please send appropriate patches to the Debian bug tracking system. OPTIONS -h print basic help -b brief - print only a summary of the statistics of the .css file (otherwise, prints a full list of how many bins are in each counter state) -q quiet mode; no warning messages -r report then exit (no menu). The default if -r is not specified is to drop into a command-menu based system. -s if no css file found, create new one with this many buckets. Default is 1 million + 1 buckets -S same as -s, but round up to next 2^n + 1 boundary. -v print version and exit -D dump css file to stdout in the architecture-independent CSV format, suitable for reloading with -R in an architecture. (note that .css files are a hardware-architecture dependent format) -R create and restore css from the hardware-architecture independent CSV format file (reads from stdin if csv-file is not supplied. THE COMMAND MENU If -r is not supplied, a menu appears with the following options. Note that all of these operations are "in place" and surgical- there is NO undo functionality. Wise users will make a backup copy of all .css files before using cssutil to alter values. -Z zero all bins at or below a value. This is useful for deleting all small-count features from the .css statistics files leaving higher- count features untouched. -S subtract a constant from all bins - this rolls all features back a constant amount. -D divide all bins by a constant - this rolls features back linearly, rather than in scalar fashion. -R rescan - regenerate the statistics output that was initially printed. -P pack - re-slot features to optimize access time. -Q - gracefully exit, saving changes. (note that since these operations are in-place and surgical, there is no option to exit without saving changes. DESCRIPTION cssutil is a general utility to manipulate and measure the .css format statistics files used by CRM114's Markovian and OSB classifiers. The biggest uses are to check the available space remaining in a .css file, to selectively groom a .css file, and to port architecture-dependent .css files to and from an ASCII CSV format, which is architecture independent. The cssutil program can be used to create information-less .css files: cssutil -b -r spam.css cssutil -b -r nonspam.css . This creates the full-size files ./spam.css and ./nonspam.css, holding no information. The cssutil program can be used check that the .css files are reasonable. Invoke cssutil as: cssutil -b -r spam.css cssutil -b -r nonspam.css You should get back a report something like this: Sparse spectra file spam.css statistics: Total available buckets : 1048576 Total buckets in use : 506987 Total hashed datums in file : 1605968 Average datums per bucket : 3.17 Maximum length of overflow chain : 39 Average length of overflow chain : 1.84 Average packing density : 0.48 Note that the packing density is 0.48; this means that this .css file is about half full of features. Once the packing density gets above about 0.9, you will notice that CRM114 will take longer to process text. The penalty is small below packing densities below about 0.95 and only about a factor of 2 at 0.97 . Best is to keep it below .7 to .8. SHORTCOMINGS Note that cssutil as of version 20040816 is NOT capable of dealing with the CRM114 Winnow classifier's floating-point .cow files. Worse, cssutil is unaware of it's shortcomings, and will try anyway. The only recourse is to be aware of this issue and not use cssutil on a Winnow classifier floating point .cow format file. HOMEPAGE AND REPORTING BUGS http://crm114.sourceforge.net/ VERSION This manpage: $Id: cssutil.azm,v 1.4 2004/08/19 09:23:24 vanbaal Exp $ This manpage describes cssutil as shipped with crm114 version 20040816.BlameClockworkOrange. AUTHOR William S. Yerazunis. Manpage typesetting by Joost van Baal and Shalendra Chhabra COPYRIGHT Copyright (C) 2001, 2002, 2003, 2004 William S. Yerazunis. This is free software, copyrighted under the FSF's GPL. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the file COPYING for more details. SEE ALSO cssmerge(1), cssdiff(1), crm(1) cssutil 20040816.BlameClockworkOrange-auto.3 19 Aug 2004 cssutil(1)