Statistiky kolem etalonu

$ cd /corp/etalon
$ldir
drwxr-xr-x  2 skoumal staff  4096 2013-05-13 19:20 compare/
drwxr-xr-x  2 skoumal staff  4096 2013-05-14 09:40 vert-done/
drwxrwxr-x  2 skoumal staff  4096 2013-05-13 15:30 vert-morf/
drwxr-xr-x  2 skoumal staff  4096 2013-05-13 18:22 vert-morf-cmp/
drwxrwxr-x  2 skoumal staff  4096 2013-05-13 15:31 vert-rules/
drwxr-xr-x  2 skoumal staff  4096 2013-05-13 19:11 vert-rules-cmp/
drwxrwxr-x  2 skoumal staff  4096 2013-05-13 12:47 vert-rules-frazrl/
drwxr-xr-x  2 skoumal staff  4096 2013-05-13 15:07 vert-rules-frazrl-cmp/
drwxrwxr-x  2 skoumal staff  4096 2013-05-13 12:47 vert-rules-frazrl-rules/
drwxr-xr-x  2 skoumal staff  4096 2013-05-13 15:07 vert-rules-frazrl-rules-cmp/

$ cd .../vert-morf
$ for ff in *; do echo $ff; normalize-anot.pl < $ff | post-disamb.pl > ../vert-morf-cmp/$ff; done

a totéž pro vert-rules, vert-rules-frazrl a vert-rules-frazrl-rules.

$ cd ../vert-done
$ for ff in *; do echo $ff; paste $ff ../vert-morf-cmp/$ff | perl -pe 's/^[\t]+$//' > ../compare/morf-$ff; done

a opět opakovat pro vert-rules, vert-rules-frazrl a vert-rules-frazrl-rules. Předpony jsou morf-, rules-, frazrl- a rfr-.

$ cd .../compare
$ cat morf-???????? | compare-Mvert tag-table.txt > /dev/null
$ cat morf-?????????? | compare-Mvert tag-table.txt > /dev/null
$ cat morf-????????? | compare-Mvert tag-table.txt > /dev/null

Z údajů Lines a Diff se spočítá procento recallu.

$ cd /corp/etalon/vert-morf
$ cat ???????? | count-spaces
$ cat ?????????? | count-spaces
$ cat ????????? | count-spaces

Zajímavé jsou pro nás údaje Forms, Tags a Uniq. Stejný výpočet provedeme i pro vert-rules, vert-rules-frazrl a vert-rules-frazrl-rules.

$ mkdir .../compare/anotatori/PUB
$ mkdir .../compare/anotatori/NOV
$ mkdir .../compare/anotatori/SCI

$ cd .../data/PUB-export
$ for ff in *; do echo $ff; prep-for-tag-comparison.sh < $ff |\
  perl -pe 's/<.*//' | cat -s > ../../compare/anotatori/PUB/$ff; done
$ cd ../NOV-export
$ for ff in *; do echo $ff; prep-for-tag-comparison.sh < $ff |\
  perl -pe 's/<.*//' | cat -s > ../../compare/anotatori/NOV/$ff; done
$ cd ../SCI-export
$ for ff in *; do echo $ff; prep-for-tag-comparison.sh < $ff |\
  perl -pe 's/<.*//' | cat -s > ../../compare/anotatori/SCI/$ff; done

$ cd .../compare/merge/done
$ for ff in ????????; do (cd ../../anotatori/PUB/; ls $ff*); done
$ cat > ../../PUB-done.txt
$  for ff in ??????????; do (cd ../../anotatori/NOV/; ls $ff*); done
$  cat > ../../NOV-done.txt
$  for ff in ?????????; do (cd ../../anotatori/SCI/; ls $ff*); done
$  cat > ../../SCI-done.txt

$ cd .../compare
$ vert-vert-vert.sh merge/archiv/<file>.fordiff anotatori/PUB/<file>-<A1> anotatori/PUB/<file>-<A2> \
  > cmp-3input/<file>-<A1>-<A2>

anebo použít skript make-3input.pl

$ make-3input.pl PUB < PUB-done.txt

make-3input.pl

 #!/usr/bin/perl
 
use autodie;
 
$num_args = $#ARGV + 1;
if ($num_args != 1) {
  print "\nUsage: make-3input.pl genre_abbrev\n";
  exit;
}
 
$genre = $ARGV[0];
 
chdir "/home/skoumal/PROJEKTY/GACR1012/compare";
 
while ($radek = <STDIN>) {
    chomp $radek;
    @file = split("[ ]+",$radek);
    @parts0 = split('-',$file[0]);
    @parts1 = split('-',$file[1]);
    print $radek;
    @inp = ("merge/archiv/$parts0[0].fordiff","anotatori/$genre/$file[0]","anotatori/$genre/$file[1]");
    $outf = "cmp-3input/$parts0[0]-$parts0[1]-$parts1[1]";
    print " > $outf\n";
    open ($outhandle,'>',$outf);
    print $outhandle `vert-vert-vert.sh $inp[0] $inp[1] $inp[2]`;
    close ($outhandle);
}

$ cd .../compare/cmp-3input
$ for ff in ????????-??-??; do echo $ff; compare-3vert ../tag-table.txt -F < $ff \
  > ../cmp-3result/$ff; done
$ cat > ../3vert-PUB-results.txt
$ for ff in ?????????-??-??; do echo $ff; compare-3vert ../tag-table.txt -F < $ff \
  > ../cmp-3result/$ff; done
$ cat > ../3vert-SCI-results.txt
$ for ff in ??????????-??-??; do echo $ff; compare-3vert ../tag-table.txt -F < $ff \
  > ../cmp-3result/$ff; done
$ cat > ../3vert-NOV-results.txt