GojolParser, a deep structure morpho-syntactic analyzer


   Has instances for English, French, German, Spanish, Italian and 
Portuguese. Tagging error rate below 1% ( e.g. for samples of Negra, 
0.2% ). Throughput for English, on a 200 MHz Pentium : 300 words/second 
for average ( i.e. below 20 words each ) sentences. Four output types : 
tree, dependency map, translational and graphic. Robustness 100% ( no 
sentence remains unparsed ). Runs on Linux, Windows ( NT, 2000, XP, 
98 ), Solaris. Older versions licenses purchased by universities and 
research institutes from Germany, Switzerland, Japan.
   Based on an innovative technology, unpublished yet. Comparisons to 
the most reputed parser on the European market proved slight superiority 
in terms of accuracy. And much lower price.
   Demo deliverable by e-mail attachment, operating 3 days (stops by 
reading the date from the Internet). 
   Author and exclusive owner of the registered property rights :

Dr. ing. Vlad Gojol
Chief Scientist, CEO
Cybernetics-Langos
Campul cu flori 6, D25, AP7
Bucharest 774082, Romania
phone: +40 21 778 5315
fax: +40 21 413 8176
e-mail: gojol@rnc.ro


1. Dependency map for an example Portuguese sentence :

O primeiro fabricante mundial de «ratos» para computador, a empresa 
suíça Logitech, apresentou esta semana numa feira especializada.

1   _O              ->3   DTAT    .DET->N          o / c=n n=s g=m 
2   primeiro        ->3   ADJC    .DET->NNCO       primeiro / c=n n=s g=m 
3   fabricante      ->11  NNCO    .PLUSN->N        fabricante / c=n n=s g=m 
4   mundial         ->3   ADJC    .ADJ->N          mundial / c=n n=s g=m 
5   de              ->3   PREP    P->N             de 
6   ratos           ->5   NNCO    .PREP->PREP      rato / c=a n=p g=m 
7   para            ->6   PREP    P->NNCO          para 
8   computador      ->7   NNCO    .PREP->PREP      computador / c=a n=s g=m 
9   ,                     COMA                     , 
10  a               ->11  DTAT    .DET->N          a / c=n n=s g=f 
11  empresa         ->15  NNCO    .SUBJ->VRBV      empresa / c=n n=s g=f 
12  suíça           ->11  ADJC    .ADJ->NNCO       suíço / c=n n=s g=f 
13  Logitech        ->11  NNPR    .GNTV->N         Logitech / c=g n=s g=m 
14  ,                     COMA                     , 
15  apresentou      root  VRBV                     apresentar / t=ipres p=3 n=s 
16  esta            ->17  DTAT    .DET->NNCO       este / c=a n=s g=f 
17  semana          ->15  NNCO    N->VRBV          semana / c=a n=s g=f 
18  numa            ->15  PREP    P->VRBV          em 
19  feira           ->18  NNCO    .PREP->PREP      feira / c=a n=s g=f 
20  especializada   ->19  VRBP    .DETVB->NNCO     especializar / c=a n=s g=f 
21  .                     STOP                     . 


2. Tree output for the same : 

(S
  (.SUBJ
    (N
      (.PLUSN
        (N
          (N
            (N
              (.DET
                (DTAT _O))
              (N
                (.DET
                  (ADJC primeiro))
                (NNCO fabricante)))
            (.ADJ
              (ADJC mundial)))
          (P
            (PREP de)
            (.PREP
              (N
                (NNCO ratos)
                (P
                  (PREP para)
                  (.PREP
                    (NNCO computador))))))))
      (COMA ,)
      (N
        (N
          (.DET
            (DTAT a))
          (N
            (NNCO empresa)
            (.ADJ
              (ADJC suíça))))
        (.GNTV
          (NNPR Logitech)))))
  (COMA ,)
  (VRBV apresentou)
  (N
    (.DET
      (DTAT esta))
    (NNCO semana))
  (P
    (PREP numa)
    (.PREP
      (N
        (NNCO feira)
        (.DETVB
          (VP
            (VRBP especializada)
            (STOP .)))))))