src/Data/Gargantext/Ngrams/Lang/Fr.hs

   1 {-# LANGUAGE OverloadedStrings #-}
   2
   3 module Data.Gargantext.Ngrams.Lang.Fr (selectNgrams, groupNgrams, textTest)
   4     where
   5
   6 import Data.Gargantext.Prelude
   7 import Data.Text (Text)
   8 import Data.Monoid ((<>))
   9
  10 selectNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  11 selectNgrams xs = pf selectNgrams' xs
  12     where
  13         selectNgrams' (_,"N"    ,_       ) = True
  14         selectNgrams' (_,"NC"   ,_       ) = True
  15         selectNgrams' (_,"NN+CC",_       ) = True
  16 -- FIXME NER in French must be improved
  17 --        selectNgrams' (_,_      ,"I-PERS") = True
  18 --        selectNgrams' (_,_      ,"I-LIEU") = True
  19         selectNgrams' (_,_      ,_       ) = False
  20
  21
  22 groupNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  23 groupNgrams []       = []
  24
  25 -- "Groupe : nom commun et adjectifs avec conjonction"
  26 groupNgrams ((n,"NC",n'):(j1,"ADJ",_):(_,"CC",_):(j2,"ADJ",_):xs) = groupNgrams (n1:n2:xs)
  27     where
  28         n1 = (n <> " " <> j1, "NC", n')
  29         n2 = (n <> " " <> j2, "NC", n')
  30
  31 -- /!\ sometimes N instead of NC (why?)
  32 groupNgrams ((n,"N",n'):(j1,"ADJ",_):(_,"CC",_):(j2,"ADJ",_):xs) = groupNgrams (n1:n2:xs)
  33     where
  34         n1 = (n <> " " <> j1, "N", n')
  35         n2 = (n <> " " <> j2, "N", n')
  36
  37 -- Groupe : Adjectif + Conjonction de coordination + Adjectif
  38 -- groupNgrams ((j1,"ADJ",_):(_,"CC",_):(j2,"ADJ",j2'):xs) = groupNgrams ((j1 <> " " <> j2, "ADJ", j2'):xs)
  39
  40 -- Groupe : Nom commun + préposition + Nom commun
  41 groupNgrams ((n1,"NC",_):(p,"P",_):(n2,"NC",n2'):xs) = groupNgrams ((n1 <> " " <> p <> " " <> n2, "NC", n2'):xs)
  42
  43 -- Groupe : Plusieurs adjectifs successifs
  44 groupNgrams ((x,"ADJ",_):(y,"ADJ",yy):xs) = groupNgrams ((x <> " " <> y, "ADJ", yy):xs)
  45
  46 -- Groupe : nom commun et adjectif
  47 groupNgrams ((x,"NC",_):(y,"ADJ",yy):xs)  = groupNgrams ((x <> " " <> y, "NC", yy):xs)
  48 -- /!\ sometimes N instead of NC (why?)
  49 groupNgrams ((x,"N",_):(y,"ADJ",yy):xs)   = groupNgrams ((x <> " " <> y, "NC", yy):xs)
  50
  51 -- Groupe : adjectif et nom commun
  52 groupNgrams ((x,"ADJ",_):(y,"NC",yy):xs)  = groupNgrams ((x <> " " <> y, "NC", yy):xs)
  53 -- /!\ sometimes N instead of NC (why?)
  54 groupNgrams ((x,"ADJ",_):(y,"N",yy):xs)   = groupNgrams ((x <> " " <> y, "NC", yy):xs)
  55
  56 -- Si aucune des règles précédentes n'est remplie
  57 groupNgrams (x:xs)                        = (x:(groupNgrams xs))
  58
  59
  60 textTest :: [String]
  61 textTest = [ "L'heure d'arrivée des coureurs dépend de la météo du jour."]
  62