src/Gargantext/Text/Lang/En.hs

   1 {-|
   2 Module      : Gargantext.Text.Lang.En
   3 Description :
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Here is a longer description of this module, containing some
  11 commentary with @some markup@.
  12 -}
  13
  14 {-# LANGUAGE NoImplicitPrelude #-}
  15 {-# LANGUAGE OverloadedStrings #-}
  16
  17 module Gargantext.Text.Lang.En (selectNgrams, groupNgrams, textTest) where
  18
  19 import Gargantext.Prelude
  20 import Data.Text (Text)
  21 import Data.Monoid ((<>))
  22
  23 selectNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  24 selectNgrams xs = filter isNgrams xs
  25     where
  26         isNgrams (_,"NN"   ,_             ) = True
  27         isNgrams (_,"NNS"  ,_             ) = True
  28         isNgrams (_,"NNP"  ,_             ) = True
  29         isNgrams (_,"NN+CC",_             ) = True
  30         isNgrams (_,_      ,"PERSON"      ) = True
  31         isNgrams (_,_      ,"ORGANIZATION") = True
  32         isNgrams (_,_      ,"LOCATION"    ) = True
  33         isNgrams (_,_      ,_             ) = False
  34
  35
  36 groupNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  37 groupNgrams []       = []
  38
  39 groupNgrams ((j1,"JJ",j1'):(c1,"CC",c1'):(j2,"JJ",j2'):(j3,"JJ",_):xs)      = groupNgrams (jn1:cc:jn2:xs)
  40     where
  41         jn j' j'' jn' = (j' <> " " <> j'', "JJ", jn')
  42         cc  = (c1, "CC", c1')
  43         jn1 = (j1, "JJ", j1')
  44         jn2 = jn j2 j3 j2'
  45
  46 groupNgrams ((j1,"JJ",_):(_,"CC",_):(j2,"JJ",_):(n,"NN",nn):xs)      = groupNgrams (jn1:jn2:xs)
  47     where
  48         jn j m mm p = (j <> " " <> m, p, mm)
  49         jn1 = jn j1 n nn ("NN+CC" :: Text)
  50         jn2 = jn j2 n nn ("NN+CC" :: Text)
  51
  52 groupNgrams ((j1,"JJ",_):(_,"CC",_):(j2,"JJ",_):(n,"NNS",nn):xs)      = groupNgrams (jn1:jn2:xs)
  53     where
  54         jn j m mm p = (j <> " " <> m, p, mm)
  55         jn1 = jn j1 n nn ("NN+CC" :: Text)
  56         jn2 = jn j2 n nn ("NN+CC" :: Text)
  57
  58 groupNgrams ((x,"JJ",_):(y,"JJ",yy):xs)    = groupNgrams ((x <> " " <> y, "JJ", yy):xs)
  59 groupNgrams ((x,"JJ",_):(y,"NN",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  60 groupNgrams ((x,"JJ",_):(y,"NNS",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  61
  62 groupNgrams ((x,"NNP",_):(y,"NN",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  63 groupNgrams ((x,"NN",_):(y,"NP",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  64
  65 groupNgrams ((x,"NN",_):(y,"NNS",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  66 groupNgrams ((x,"NP",_):(y,"NP",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  67
  68 groupNgrams ((x,"NN",_):(y,"NN",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  69
  70
  71 -- extractNgrams "Test the antiinflammatory or analgesic activity?"
  72 -- [[("``","``","O"),("Test","VB","O"),("the","DT","O"),("antiinflammatory activity analgesic activity","NN","O"),("?",".","O"),("''","''","O")]]
  73 -- > should be (antiinflammatory activity) <> (analgesic activity)
  74
  75 groupNgrams ((x,"NN",_):(o,"IN",_):(y,"NN",yy):xs)       = groupNgrams ((x <> " " <> o <> " " <> y, "NN", yy):xs)
  76 groupNgrams ((x,"NN",_):(o,"IN",_):(y,"NNP",yy):xs)      = groupNgrams ((x <> " " <> o <> " " <> y, "NN", yy):xs)
  77
  78 groupNgrams ((x,"NN",_):(o,"IN",_):(det,"DT",_):(y,"NN",yy):xs)    = groupNgrams ((x <> " " <> o <> " " <> det <> " " <> y, "NN", yy):xs)
  79 groupNgrams ((x,"NN",_):(o,"IN",_):(det,"DT",_):(y,"NNP",yy):xs)   = groupNgrams ((x <> " " <> o <> " " <> det <> " " <> y, "NN", yy):xs)
  80
  81 groupNgrams ((x,_,"PERSON"):(y,yy,"PERSON"):xs)             = groupNgrams ((x <> " " <> y,yy,"PERSON"):xs)
  82 groupNgrams ((x,_,"ORGANIZATION"):(y,yy,"ORGANIZATION"):xs) = groupNgrams ((x <> " " <> y,yy,"ORGANIZATION"):xs)
  83 groupNgrams ((x,_,"LOCATION"):(y,yy,"LOCATION"):xs)         = groupNgrams ((x <> " " <> y,yy,"LOCATION"):xs)
  84
  85 groupNgrams (x:xs)                                          = (x:(groupNgrams xs))
  86
  87
  88 textTest :: [Text]
  89 textTest = [ "Alcoholic extract of Kaempferia galanga was tested for analgesic and antiinflammatory activities in animal models. "
  90            , "Three doses, 300 mg/kg, 600 mg/kg and 1200 mg/kg of the plant extract prepared as a suspension in 2 ml of 2% gum acacia were used. "
  91            , " Acute and sub acute inflammatory activities were studied in rats by carrageenan induced paw edema and cotton pellet induced granuloma models respectively. "
  92           , "In both models, the standard drug used was aspirin 100 mg/kg. "
  93           , "Two doses 600 mg/kg and 1200 mg/kg of plant extract exhibited significant (P<0.001) antiinflammatory activity in carrageenan model and cotton pellet granuloma model in comparison to control. "
  94          , "Analgesic activity was studied in rats using hot plate and tail-flick models. "
  95          , "Codeine 5 mg/kg and vehicle served as standard and control respectively. "
  96          , "The two doses of plant extract exhibited significant analgesic activity in tail flick model (P<0.001) and hot plate model (P<0.001) in comparison to control. "
  97          , "In conclusion K. galanga possesses antiinflammatory and analgesic activities. "]
  98
  99