


CMAT(1)                  USER COMMANDS                    CMAT(1)



NAME
     cmat - produce a confusion matrix

SYNOPSIS
     cmat (-I) (-l lin_len) (-s) (result_file)

DESCRIPTION
     cmat is a program to construct a  confusion  matrix  from  a
     data  file containing the results of a recognition test. The
     input consists of a sequence of pairs of names, of which the
     first is the name by which a token should be recognised, and
     the second the name by which it  has  actually  been  recog-
     nised.   Names,  whether  in  the  same  pair or in adjacent
     pairs, can be separated by any combination of  spaces,  tabs
     or  "new  line"s. Names will be listed in the order in which
     they first appear as the first element of a pair.  The  pro-
     gram  gives  recognition  scores as a percentage; if a given
     slot in the matrix corresponds to no recognitions at all, it
     is  left  blank;  if  a  "0"  is printed, it means that some
     recognitions occurred for that slot, but  amounted  to  less
     than  1%.  The confusion matrix will be written to the stan-
     dard output; if no result file is specified, input  will  be
     taken from the standard input.

     Options and their meanings are:

     -I         Identify program and exit

     -lline_length
                Specify number of characters across a line of the
                output matrix; default 70

     -s         Specify that token  names  will  be  on  adjacent
                lines  (default  is  double  line spacing between
                token names)

VERSION/AUTHOR
     2.04 - Peter Davies (based on earlier program by B. L. Tan)

SOURCE
     /users/mark/sfs/man/../ucl/cmat.c














SPAR                    Last change: UCL                        1



