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