inference.pl: solve for final legitimate output
usage: inference.pl [options]

Required options
  -b   <filename>        boundary input
  -r   <filename>        probability table input
  -A   <filename>        argument output

  -c   <filename>        corpus input
  -v   <filename>        verb input
                         If not specified, the final arugment predictions
                         will assume one verb per sentence.
                         Only one of -c or -v is required.

  -i   <filename>        illegal arugment list

Optional options
  -w   <filename>        Input text
  -W                     Output text column before verb column

  -p   <filename>        parse tree input
  -P   <filename>        parse tree output

  -V   <filename>        verb output
                         The verbs as input will be directly copied to the
                         verb ouptut.  This is useful in the later stages of
                         SRL if it is done through pipe.  The default is not
                         not to read and copy the verbs.

  -C   <filename>        corpus output
                         The content is exactly the same as the corpus input.
                         This can be useful for the later stages of SRL.  If
                         -C is not specified, the corpus will not be output.

  -B   <filename>        boundary output
                         The content is exactly the same as the boundary input.
                         If -B is not specified, the output is not generated.