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

   1 {-# LANGUAGE OverloadedStrings #-}
   2
   3 module Data.Gargantext.Ngrams.Lang.En (selectNgrams, groupNgrams, textTest) where
   4
   5 import Data.Gargantext.Prelude
   6 import Data.Text (Text)
   7 import Data.Monoid ((<>))
   8
   9 selectNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  10 selectNgrams xs = pf selectNgrams' xs
  11     where
  12         selectNgrams' (_,"NN"   ,_             ) = True
  13         selectNgrams' (_,"NNS"  ,_             ) = True
  14         selectNgrams' (_,"NNP"  ,_             ) = True
  15         selectNgrams' (_,"NN+CC",_             ) = True
  16         selectNgrams' (_,_      ,"PERSON"      ) = True
  17         selectNgrams' (_,_      ,"ORGANIZATION") = True
  18         selectNgrams' (_,_      ,"LOCATION"    ) = True
  19         selectNgrams' (_,_      ,_             ) = False
  20
  21
  22 groupNgrams :: [(Text, Text, Text)] -> [(Text, Text, Text)]
  23 groupNgrams []       = []
  24
  25 groupNgrams ((j1,"JJ",j1'):(c1,"CC",c1'):(j2,"JJ",j2'):(j3,"JJ",_):xs)      = groupNgrams (jn1:cc:jn2:xs)
  26     where
  27         jn j' j'' jn' = (j' <> " " <> j'', "JJ", jn')
  28         cc  = (c1,"CC",c1')
  29         jn1 = (j1, "JJ", j1')
  30         jn2 = jn j2 j3 j2'
  31
  32 groupNgrams ((j1,"JJ",_):(_,"CC",_):(j2,"JJ",_):(n,"NN",nn):xs)      = groupNgrams (jn1:jn2:xs)
  33     where
  34         jn j m mm p = (j <> " " <> m, p, mm)
  35         jn1 = jn j1 n nn ("NN+CC" :: Text)
  36         jn2 = jn j2 n nn ("NN+CC" :: Text)
  37
  38 groupNgrams ((j1,"JJ",_):(_,"CC",_):(j2,"JJ",_):(n,"NNS",nn):xs)      = groupNgrams (jn1:jn2:xs)
  39     where
  40         jn j m mm p = (j <> " " <> m, p, mm)
  41         jn1 = jn j1 n nn ("NN+CC" :: Text)
  42         jn2 = jn j2 n nn ("NN+CC" :: Text)
  43
  44 groupNgrams ((x,"JJ",_):(y,"JJ",yy):xs)    = groupNgrams ((x <> " " <> y, "JJ", yy):xs)
  45 groupNgrams ((x,"JJ",_):(y,"NN",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  46 groupNgrams ((x,"JJ",_):(y,"NNS",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  47
  48 groupNgrams ((x,"NNP",_):(y,"NN",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  49 groupNgrams ((x,"NN",_):(y,"NP",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  50
  51 groupNgrams ((x,"NN",_):(y,"NNS",yy):xs)   = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  52 groupNgrams ((x,"NP",_):(y,"NP",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  53
  54 groupNgrams ((x,"NN",_):(y,"NN",yy):xs)    = groupNgrams ((x <> " " <> y, "NN", yy):xs)
  55
  56
  57 -- extractNgrams "Test the antiinflammatory or analgesic activity?"
  58 -- [[("``","``","O"),("Test","VB","O"),("the","DT","O"),("antiinflammatory activity analgesic activity","NN","O"),("?",".","O"),("''","''","O")]]
  59 -- > should be (antiinflammatory activity) <> (analgesic activity)
  60
  61 groupNgrams ((x,"NN",_):(o,"IN",_):(y,"NN",yy):xs)       = groupNgrams ((x <> " " <> o <> " " <> y, "NN", yy):xs)
  62 groupNgrams ((x,"NN",_):(o,"IN",_):(y,"NNP",yy):xs)       = groupNgrams ((x <> " " <> o <> " " <> y, "NN", yy):xs)
  63
  64 groupNgrams ((x,"NN",_):(o,"IN",_):(det,"DT",_):(y,"NN",yy):xs)       = groupNgrams ((x <> " " <> o <> " " <> det <> " " <> y, "NN", yy):xs)
  65 groupNgrams ((x,"NN",_):(o,"IN",_):(det,"DT",_):(y,"NNP",yy):xs)       = groupNgrams ((x <> " " <> o <> " " <> det <> " " <> y, "NN", yy):xs)
  66
  67 groupNgrams ((x,_,"PERSON"):(y,yy,"PERSON"):xs)             = groupNgrams ((x <> " " <> y,yy,"PERSON"):xs)
  68 groupNgrams ((x,_,"ORGANIZATION"):(y,yy,"ORGANIZATION"):xs) = groupNgrams ((x <> " " <> y,yy,"ORGANIZATION"):xs)
  69 groupNgrams ((x,_,"LOCATION"):(y,yy,"LOCATION"):xs) = groupNgrams ((x <> " " <> y,yy,"LOCATION"):xs)
  70
  71 groupNgrams (x:xs)                                          = (x:(groupNgrams xs))
  72
  73
  74 textTest :: [String]
  75 textTest = [ "Alcoholic extract of Kaempferia galanga was tested for analgesic and antiinflammatory activities in animal models. "
  76            , "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. "
  77            , " Acute and sub acute inflammatory activities were studied in rats by carrageenan induced paw edema and cotton pellet induced granuloma models respectively. "
  78           , "In both models, the standard drug used was aspirin 100 mg/kg. "
  79           , "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. "
  80          , "Analgesic activity was studied in rats using hot plate and tail-flick models. "
  81          , "Codeine 5 mg/kg and vehicle served as standard and control respectively. "
  82          , "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. "
  83          , "In conclusion K. galanga possesses antiinflammatory and analgesic activities. "]
  84
  85