src/Gargantext/Core/Ext/IMT.hs

   1 {-|
   2 Module      : Gargantext.API
   3 Description : Server API
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9 -}
  10
  11
  12 module Gargantext.Core.Ext.IMT where
  13
  14 import Data.Either (Either(..))
  15 import Data.Map (Map)
  16 import Data.Text (Text, splitOn)
  17
  18 import qualified Data.Set    as S
  19 import qualified Data.List   as DL
  20 import qualified Data.Vector as DV
  21 import qualified Data.Map    as M
  22 import qualified Prelude
  23
  24 import Gargantext.Prelude
  25
  26 import Gargantext.Core.Text.Metrics.Utils      as Utils
  27 import Gargantext.Core.Text.Corpus.Parsers.CSV as CSV
  28
  29 data School = School { school_shortName :: Text
  30                      , school_longName  :: Text
  31                      , school_id        :: Text
  32 } deriving (Show, Read, Eq)
  33
  34 schools :: [School]
  35 schools = [ School
  36             { school_shortName = "Mines Albi-Carmaux"
  37             , school_longName = "Mines Albi-Carmaux - École nationale supérieure des Mines d'Albi‐Carmaux"
  38             , school_id = "469216" }
  39           , School
  40             { school_shortName = "Mines Alès"
  41             , school_longName = "EMA - École des Mines d'Alès"
  42             , school_id = "6279" }
  43           , School
  44             { school_shortName = "Mines Douai"
  45             , school_longName = "Mines Douai EMD - École des Mines de Douai"
  46             , school_id = "224096" }
  47           , School
  48             { school_shortName = "Mines Lille"
  49             , school_longName = "Mines Lille - École des Mines de Lille"
  50             , school_id = "144103" }
  51           , School
  52             { school_shortName = "IMT Lille Douai"
  53             , school_longName = "IMT Lille Douai"
  54             , school_id = "497330" }
  55           , School
  56             { school_shortName = "Mines Nantes"
  57             , school_longName = "Mines Nantes - Mines Nantes"
  58             , school_id = "84538" }
  59           , School
  60             { school_shortName = "Télécom Bretagne"
  61             , school_longName = "Télécom Bretagne"
  62             , school_id = "301262" }
  63           , School
  64             { school_shortName = "IMT Atlantique"
  65             , school_longName = "IMT Atlantique - IMT Atlantique Bretagne-Pays de la Loire"
  66             , school_id = "481355" }
  67           , School
  68             { school_shortName = "Mines Saint-Étienne"
  69             , school_longName = "Mines Saint-Étienne MSE - École des Mines de Saint-Étienne"
  70             , school_id = "29212" }
  71           , School
  72             { school_shortName = "Télécom École de Management"
  73             , school_longName = "TEM - Télécom Ecole de Management"
  74             , school_id = "301442" }
  75           , School
  76             { school_shortName = "IMT Business School"
  77             , school_longName = "IMT Business School"
  78             , school_id = "542824" }
  79           , School
  80             { school_shortName = "Télécom ParisTech"
  81             , school_longName = "Télécom ParisTech"
  82             , school_id = "300362" }
  83           , School
  84             { school_shortName = "Télécom SudParis"
  85             , school_longName = "TSP - Télécom SudParis"
  86             , school_id = "352124" }
  87           , School
  88             { school_shortName = "ARMINES"
  89             , school_longName = "ARMINES"
  90             , school_id = "300362" }
  91           , School
  92             { school_shortName = "Eurecom"
  93             , school_longName = "Eurecom"
  94             , school_id = "421532" }
  95           , School
  96             { school_shortName = "Mines ParisTech"
  97             , school_longName = "MINES ParisTech - École nationale supérieure des mines de Paris"
  98             , school_id = "301492" }
  99             ]
 100
 101 mapIdSchool :: Map Text Text
 102 mapIdSchool = M.fromList $ Gargantext.Prelude.map
 103                 (\(School { school_shortName, school_id }) -> (school_id, school_shortName)) schools
 104
 105 hal_data :: IO (Either Prelude.String (DV.Vector CsvHal))
 106 hal_data = do
 107   r <- CSV.readCsvHal "doc/corpus_imt/Gargantext_Corpus.csv"
 108   pure $ snd <$> r
 109
 110 names :: S.Set Text
 111 names = S.fromList $ Gargantext.Prelude.map (\s -> school_id s) schools
 112
 113 toSchoolName :: Text -> Text
 114 toSchoolName t = case M.lookup t mapIdSchool of
 115   Nothing -> t
 116   Just t' -> t'
 117
 118 publisBySchool :: DV.Vector CsvHal -> [(Maybe Text, Int)]
 119 publisBySchool hal_data' = Gargantext.Prelude.map (\(i,n) -> (M.lookup i mapIdSchool, n))
 120                         $ DL.filter (\i -> S.member  (fst i) names)
 121                         $ DL.reverse
 122                         $ DL.sortOn snd
 123                         $ M.toList
 124                         $ Utils.freq
 125                         $ DL.concat
 126                         $ DV.toList
 127                         $ DV.map (\n -> splitOn ( ", ") (csvHal_instStructId_i n) )
 128                         $ DV.filter (\n -> csvHal_publication_year n == 2017) hal_data'
 129
 130