src/Data/Gargantext/Ngrams/Hetero_hs

   1 module Data.Gargantext.Ngrams.Hetero where
   2
   3 import GHC.Real as R
   4 import Data.Set as S
   5 import Data.Map as M
   6 import Data.List.Split as S
   7 import Database.PostgreSQL.Simple as PGS
   8 import Opaleye.PGTypes (PGInt4)
   9 import Opaleye.Internal.Column (Column)
  10
  11 import Data.Gargantext.Database.Gargandb
  12 import Data.Gargantext.Database.Private
  13 --import Data.Gargantext.Utils.Chronos
  14
  15 import Data.Gargantext.Ngrams.Words (cleanText)
  16 import Data.Gargantext.Ngrams.Count (occurrences)
  17
  18 import Data.Gargantext.Database.Simple
  19
  20 --main = do
  21 --    t <- getTextquery
  22 --    print (Prelude.map (heterogeinity . concat) $ S.chunksOf 3 t)
  23
  24 -- heterogeinity sur concat texts
  25 heterogeinity' :: Int -> Int -> Int -> IO [Integer]
  26 heterogeinity' corpus_id limit x = do
  27     t <- getAbstract corpus_id limit
  28     Prelude.mapM (dicoStruct . occurrences) $ (S.chunksOf x) .  cleanText $ concat t
  29
  30 heterogeinity'' :: Int -> Int -> Int -> IO [Integer]
  31 heterogeinity'' corpus_id limit size = do
  32     t <- getAbstract corpus_id limit
  33     Prelude.mapM (dicoStruct . occurrences) $ (S.chunksOf size) .  cleanText $ concat t
  34
  35
  36 dicoStruct :: (Integral r, Monad m) => M.Map t r -> m r
  37 dicoStruct dict_occ = do
  38     let keys_size = toInteger $ length $ M.keys dict_occ
  39     let total_occ = sum $ Prelude.map (\(x, y) -> y) $ M.toList dict_occ
  40     return $ div total_occ (fromIntegral keys_size)
  41
  42 -- heterogeinity sur UCT (Unité de Context Textuel)
  43 heterogeinity :: [Char] -> IO Integer
  44 heterogeinity string = do
  45     let dict_occ = occurrences $ cleanText string
  46
  47     let keys_size = toInteger $ length $ M.keys dict_occ
  48     let total_occ = sum $ Prelude.map (\(x, y) -> y) $ M.toList dict_occ
  49
  50     return $ div total_occ (fromIntegral keys_size)
  51
  52
  53 --computeHeterogeinity
  54 --  :: Fractional t =>
  55 --       Opaleye.Internal.Column.Column Opaleye.PGTypes.PGInt4
  56 --            -> IO (t, Integer, Integer)
  57 computeHeterogeinity corpus_id = do
  58     c <- PGS.connect infoGargandb
  59     t <- getText c (nodeHyperdataText corpus_id)
  60     heterogeinity $ Prelude.concat t
  61
  62 main2 = do
  63     let corpus_ids = [
  64                 ("ALL", 272927) -- 73
  65                ,("Histoire", 1387736) -- 28
  66                ,("Sciences Po", 1296892) -- 37
  67                ,("Phylosophie", 1170004) -- 20
  68                ,("Psychologie", 1345852) -- 37
  69                ,("Sociologie", 1246452)  -- 42
  70                ]
  71
  72     r <- Prelude.map computeHeterogeinity $ Prelude.map (\(t,id) -> id) corpus_ids
  73     return r
  74
  75