src/Gargantext/Text/Metrics.hs

   1 {-|
   2 Module      : Gargantext.Text.Metrics
   3 Description : All parsers of Gargantext in one file.
   4 Copyright   : (c) CNRS, 2017 - present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Mainly reexport functions in @Data.Text.Metrics@
  11 -}
  12
  13 {-# LANGUAGE NoImplicitPrelude #-}
  14
  15 module Gargantext.Text.Metrics where
  16
  17 import Data.Text (Text, pack)
  18 import Data.List (concat)
  19
  20 --import GHC.Real (Ratio)
  21 --import qualified Data.Text.Metrics as DTM
  22
  23 import Gargantext.Prelude
  24
  25 import Gargantext.Text.Metrics.Occurrences (occurrences, cooc)
  26 import Gargantext.Text.Terms (TermType(Multi), terms)
  27 import Gargantext.Core (Lang(EN))
  28 import Gargantext.Core.Types (Terms(..))
  29 import Gargantext.Text.Context (splitBy, SplitContext(Sentences))
  30
  31 --noApax :: Ord a => Map a Occ -> Map a Occ
  32 --noApax m = M.filter (>1) m
  33
  34
  35 metrics_text :: Text
  36 metrics_text = "A table is an object. A glas is an object. The glas is on the table. The spoon is an object. The spoon is on the table."
  37
  38 -- | Sentences
  39 metrics_sentences :: [Text]
  40 metrics_sentences = [ "A table is an object."
  41                     , "A glas is an object."
  42                     , "The glas is on the table."
  43                     , "The spoon is an object."
  44                     , "The spoon is on the table."
  45                     ]
  46
  47
  48 metrics_sentences_Test = splitBy (Sentences 0) metrics_text == metrics_sentences
  49
  50 -- | Terms reordered to visually check occurrences
  51 metrics_terms :: [[[Text]]]
  52 metrics_terms = [[["table"],["object"]                  ]
  53                 ,[          ["object"],["glas"]         ]
  54                 ,[["table"],           ["glas"]         ]
  55                 ,[          ["object"],        ["spoon"]]
  56                 ,[["table"],                   ["spoon"]]
  57                 ]
  58 --metrics_terms_Test = (mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text) == metrics_terms
  59
  60 -- | Occurrences
  61 {-
  62 fromList [ (fromList ["table"] ,fromList [(["table"] , 3 )])]
  63          , (fromList ["object"],fromList [(["object"], 3 )])
  64          , (fromList ["glas"]  ,fromList [(["glas"]  , 2 )])
  65          , (fromList ["spoon"] ,fromList [(["spoon"] , 2 )])
  66 -}
  67 metrics_occ = occurrences <$> concat <$> (mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text)
  68
  69 {-
  70 -- fromList [((["glas"],["object"]),6)
  71             ,((["glas"],["spoon"]),4)
  72             ,((["glas"],["table"]),6),((["object"],["spoon"]),6),((["object"],["table"]),9),((["spoon"],["table"]),6)]
  73
  74 -}
  75 metrics_cooc = cooc <$> (mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text)
  76
  77 metrics_cooc' = (mapM (terms Multi EN) $ splitBy (Sentences 0) "The table object. The table object.")
  78
  79
  80
  81
  82