## tagset description used to compute short tag versions
TAGSET ../../tagset.dat


## features extracted for all words
RULES ALL

  funct@     [0,0]   w in_set ../data/function.dat
  unk@       [0,0]   na is 0
  possib_NA@ [0,0]   pt matches ^[NA]

  cap@       [-4,4]  W matches ^[[:upper:]][[:lower:]]
  punct@     [-3,1]  pt matches ^F
  is_dot@    [-3,1]  W matches ^[\.\?\!]$
  has_dot@   [-2,2]  W matches \.
  no_analysis@    [-3,3]    na is 0     
  all_caps@       [-2,2]    W matches ^[[:upper:]][[:upper:]]+$

  ini@            [-2,2]    W matches ^[[:upper:]]\.?$
  acr@            [-2,2]    W matches ^[[:upper:]][\.[:upper:]]+$

  rom@            [-2,2]    W matches ^[VIX]+$
  has_num@        [-2,2]    W matches [[:digit:]]
  is_num@         [-2,2]    W matches ^[[:digit:]\.\,\?]+$
  guio@           [-2,2]    W matches ^[[:alpha:]]+\-
  subgio@         [-2,2]    W matches ^[[:alpha:]]+\_
  web@            [-2,2]    W matches ^www\.*\.[[:alpha:]]+$
  auxw@           [-4,4]    w in_set ../data/function.dat
  
  gaz@            [-3,3]     pl in_set ../data/gazPER-c.dat
  gaz@            [-3,3]     pl in_set ../data/gazLOC-c.dat
  gaz@            [-3,3]     pl in_set ../data/gazORG-c.dat
  tw@             [-4,2]     pl.split("_") in_set ../data/twPER.dat
  tw@             [-4,2]     pl.split("_") in_set ../data/twLOC.dat
  tw@             [-4,2]     pl.split("_") in_set ../data/twORG.dat
  tw@             [-4,2]     pl.split("_") in_set ../data/twGENT.dat

  f@:$w(0)             [-4,4]    ALL
  pl@:$pl(0)           [-4,4]    ALL
  pt@:$pt(0)           [-3,3]    ALL
  bw@:$w(0)_$w(1)      [-3,3]    ALL
  pbt@:$pt(0)_$pt(1)   [-3,3]    ALL

  bow:$w(0)            [-6,-1]   ALL
  bow:$w(0)            [1,6]     ALL
  bol:$pl(0)           [-6,-1]   ALL
  bol:$pl(0)           [1,6]     ALL
  pa@:$pa(0)           [-3,3]    ALL

  wpt@:$w(0)_$pt(0)    [-3,3]    ALL
  plpt@:$pl(0)_$pt(0)  [-3,3]    ALL

  cap&pnoun@        [-3,3]   W matches ^[[:upper:]] AND pt matches ^N
  cap&na=0@         [-3,3]   W matches ^[[:upper:]] AND na is 0
  
  rquot:$w(0)_"    [-5,-2]    w is "
  lquot:"_$w(0)    [2,5]      w is "
  
  suf3@:{$0}     [-3,3]  w matches ...$
  pre3@:{$0}     [-3,3]  w matches ^...
  suf4@:{$0}     [-3,3]  w matches ....$
  pre4@:{$0}     [-3,3]  w matches ^....

  nw@:{nwords(0)}    [-3,3] ALL
  {quoted(0)}        [0,0]  ALL
  {parenthesis(0)}   [0,0]  ALL  

  p0@:{pattern(0)}  [0,0]  ALL
  p1L:{pattern(-1)}_{pattern(0)}                             [0,0] ALL
  p2L:{pattern(-2)}_{pattern(-1)}_{pattern(0)}               [0,0] ALL
  p3L:{pattern(-3)}_{pattern(-2)}_{pattern(-1)}_{pattern(0)} [0,0] ALL
  p1R:{pattern(0)}_{pattern(1)}                              [0,0] ALL
  p2R:{pattern(0)}_{pattern(1)}_{pattern(2)}                 [0,0] ALL
  p3R:{pattern(0)}_{pattern(1)}_{pattern(2)}_{pattern(3)}    [0,0] ALL

ENDRULES