src/Gargantext/Text/Metrics/FrequentItemSet.hs

   1 {-|
   2 Module      : Gargantext.Text.Metrics.FrequentItemSet
   3 Description : Ngrams tools
   4 Copyright   : (c) CNRS, 2018
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Domain Specific Language to manage Frequent Item Set (FIS)
  11
  12 -}
  13
  14 {-# LANGUAGE NoImplicitPrelude #-}
  15
  16 module Gargantext.Text.Metrics.FrequentItemSet
  17   ( Fis, Size(..)
  18   , occ_hlcm, cooc_hlcm
  19   , all, between
  20   , fisWithSize
  21   , fisWith
  22   , fisWithSizePoly
  23   , fisWithSizePoly2
  24   , module HLCM
  25   )
  26   where
  27
  28 import Prelude (Functor(..)) -- TODO
  29 import qualified Data.Map.Strict as Map
  30 import Data.Map.Strict (Map)
  31 import qualified Data.Set as Set
  32 import Data.Set (Set)
  33 import qualified Data.Vector as V
  34 import Data.Vector (Vector)
  35
  36 import Data.List (filter, concat)
  37 import Data.Maybe (catMaybes)
  38
  39 import HLCM
  40
  41 import Gargantext.Prelude
  42
  43 data Size = Point Int | Segment Int Int
  44
  45 ------------------------------------------------------------------------
  46 -- | Occurrence is Frequent Item Set of size 1
  47 occ_hlcm  :: Frequency -> [[Item]] -> [Fis]
  48 occ_hlcm = fisWithSize (Point 1)
  49
  50 -- | Cooccurrence is Frequent Item Set of size 2
  51 cooc_hlcm :: Frequency -> [[Item]] -> [Fis]
  52 cooc_hlcm = fisWithSize (Point 2)
  53
  54 all :: Frequency -> [[Item]] -> [Fis]
  55 all = fisWith Nothing
  56
  57 ------------------------------------------------------------------------
  58 between :: (Int, Int) -> Frequency -> [[Item]] -> [Fis]
  59 between (x,y) = fisWithSize (Segment x y)
  60
  61 --maximum :: Int -> Frequency -> [[Item]] -> [Fis]
  62 --maximum m = between (0,m)
  63
  64
  65 ------------------------------------------------------------------------
  66 ------------------------------------------------------------------------
  67 -- | Data type to type the Frequent Item Set
  68 -- TODO replace List with Set in fisItemSet
  69 -- be careful : risks to erase HLCM behavior
  70 type Fis = Fis' Item
  71 data Fis' a = Fis' { _fisCount   :: Int
  72                    , _fisItemSet :: [a]
  73                    } deriving (Show)
  74
  75 instance Functor Fis' where
  76   fmap f (Fis' c is) = Fis' c (fmap f is)
  77
  78 -- | Sugar from items to FIS
  79 items2fis :: [Item] -> Maybe Fis
  80 items2fis []     = Nothing
  81 items2fis (i:is) = Just $ Fis' i is
  82
  83 ------------------------------------------------------------------------
  84 ------------------------------------------------------------------------
  85
  86 fisWithSize :: Size -> Frequency -> [[Item]] -> [Fis]
  87 fisWithSize n f is = case n of
  88   Point  n'   -> fisWith (Just (\x -> length x == (n'+1) )) f is
  89   Segment a b -> fisWith (Just (\x -> cond a (length x) b)) f is
  90                     where
  91                       cond a' x b' = a' <= x && x <= b'
  92
  93
  94 fisWith :: Maybe ([Item] -> Bool) -> Frequency -> [[Item]] -> [Fis]
  95 fisWith s f is = catMaybes $ map items2fis $ filter' $ runLCMmatrix is f
  96   where
  97     filter' = case s of
  98                 Nothing  -> identity
  99                 Just fun -> filter fun
 100
 101 -- Here the sole purpose to take the keys as a Set is tell we do not want
 102 -- duplicates.
 103 fisWithSizePoly :: Ord a => Size -> Frequency -> Set a -> [[a]] -> [Fis' a]
 104 fisWithSizePoly n f ks = map (fmap fromItem) . fisWithSize n f . map (map toItem)
 105   where
 106     ksv = V.fromList $ Set.toList ks
 107     ksm = Map.fromList . flip zip [0..] $ V.toList ksv
 108     toItem = (ksm Map.!)
 109     fromItem = (ksv V.!)
 110
 111 fisWithSizePoly2 :: Ord a => Size -> Frequency -> [[a]] -> [Fis' a]
 112 fisWithSizePoly2 n f is = fisWithSizePoly n f ks is
 113   where
 114     ks = Set.fromList $ concat is
 115
 116
 117 ------------------------------------------------------------------------
 118 ------------------------------------------------------------------------
 119
 120
 121 --
 122 ---- | /!\ indexes are not the same:
 123 --
 124 ---- | Index ngrams from Map
 125 ----indexNgram :: Ord a => Map a Occ -> Map Index a
 126 ----indexNgram m = fromList (zip [1..] (keys m))
 127 --
 128 ---- | Index ngrams from Map
 129 ----ngramIndex :: Ord a => Map a Occ -> Map a Index
 130 ----ngramIndex m = fromList (zip (keys m) [1..])
 131 --
 132 --indexWith :: Ord a => Map a Occ -> [a] -> [Int]
 133 --indexWith m xs = unMaybe $ map (\x -> lookupIndex x m) xs
 134 --
 135 --indexIt :: Ord a => [[a]] -> (Map a Int, [[Int]])
 136 --indexIt xs = (m, is)
 137 --  where
 138 --    m  = sumOcc (map occ  xs)
 139 --    is = map    (indexWith m) xs
 140 --
 141 --list2fis :: Ord a => FIS.Frequency -> [[a]] -> (Map a Int, [FIS.Fis])
 142 --list2fis n xs = (m', fs)
 143 --  where
 144 --    (m, is) = indexIt xs
 145 --    m'      = M.filter (>50000) m
 146 --    fs      = FIS.all n is
 147 --
 148 --text2fis :: FIS.Frequency -> [Text] -> (Map Text Int, [FIS.Fis])
 149 --text2fis n xs = list2fis n (map terms xs)
 150 --
 151 ----text2fisWith :: FIS.Size -> FIS.Frequency -> [Text] -> (Map Text Int, [FIS.Fis])
 152 ----text2fisWith = undefined
 153 --
 154