src/Data/Gargantext/Ngrams/TextMining.hs

   1 module Data.Gargantext.Ngrams.TextMining where
   2
   3 import Data.Map (empty, Map, insertWith, toList)
   4 import Data.List (foldl, foldl')
   5 import qualified Data.List as L
   6
   7
   8 sortGT (a1, b1) (a2, b2)
   9         | a1 < a2 = GT
  10         | a1 > a2 = LT
  11         | a1 == a2 = compare b1 b2
  12
  13
  14 --histogram :: Ord a => [a] -> [(a, Int)]
  15 --histogram = map (head &&& length) Prelude.. group Prelude.. sort Prelude.. words
  16 --histogram = sortGT Prelude.. $ map (head &&& length) Prelude.. group Prelude.. sort Prelude.. words
  17
  18 countElem :: (Ord k) => Data.Map.Map k Int -> k -> Data.Map.Map k Int
  19 countElem m e = Data.Map.insertWith (\n o -> n + o) e 1 m
  20
  21 freqList :: (Ord k) => [k] -> Data.Map.Map k Int
  22 freqList = foldl countElem Data.Map.empty
  23
  24 --getMaxFromMap :: Data.Map.Map -> Maybe -> [a] -> [a]
  25 getMaxFromMap m = go [] Nothing (toList m)
  26   where
  27     go ks _        []           = ks
  28     go ks Nothing  ((k,v):rest) = go (k:ks) (Just v) rest
  29     go ks (Just u) ((k,v):rest)
  30         | v < u     = go ks     (Just u) rest
  31         | v > u     = go [k]    (Just v) rest
  32         | otherwise = go (k:ks) (Just v) rest
  33
  34 merge :: [a] -> [a] -> [a]
  35 merge [] ys = ys
  36 merge (x:xs) ys = x:merge ys xs
  37
  38 average :: [Double] -> Double
  39 average x = L.sum x / L.genericLength x
  40
  41 average' :: [Int] -> Double
  42 average' x = (L.sum y) / (L.genericLength y) where
  43     y = map fromIntegral x
  44
  45
  46 countYear :: [Integer] -> Map Integer Integer
  47 countYear []  = empty
  48 countYear (x:xs) = insertWith (+) x 1 (countYear xs)
  49
  50 countYear' :: [Integer] -> Map Integer Integer
  51 countYear' (xs) = foldl' (\x y -> insertWith (+) y 1 x) empty xs
  52
  53
  54 textMiningMain :: IO ()
  55 textMiningMain = do
  56         print $ merge ["abc"] ["bcd"]