src/Gargantext/Core/Methods/Similarities/Accelerate/Conditional.hs

   1 {-|
   2 Module      : Gargantext.Core.Methods.Similarities.Accelerate.Conditional
   3 Description :
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 This module aims at implementig distances of terms context by context is
  11 the same referential of corpus.
  12
  13 Implementation use Accelerate library which enables GPU and CPU computation
  14 See Gargantext.Core.Methods.Graph.Accelerate)
  15
  16 -}
  17
  18 {-# LANGUAGE TypeFamilies        #-}
  19 {-# LANGUAGE TypeOperators       #-}
  20 {-# LANGUAGE ScopedTypeVariables #-}
  21 {-# LANGUAGE ViewPatterns        #-}
  22
  23 module Gargantext.Core.Methods.Similarities.Accelerate.Conditional
  24   where
  25
  26 -- import qualified Data.Foldable as P (foldl1)
  27 -- import Debug.Trace (trace)
  28 import Data.Array.Accelerate
  29 import Data.Array.Accelerate.Interpreter (run)
  30 import Gargantext.Core.Methods.Matrix.Accelerate.Utils
  31 import Gargantext.Core.Methods.Similarities.Accelerate.SpeGen
  32 import qualified Gargantext.Prelude as P
  33
  34
  35 -- * Metrics of proximity
  36 -----------------------------------------------------------------------
  37 -- ** Conditional distance
  38
  39 -- *** Conditional distance (basic)
  40
  41 -- | Conditional distance (basic version)
  42 --
  43 -- 2 main metrics are actually implemented in order to compute the
  44 -- proximity of two terms: conditional and distributional
  45 --
  46 -- Conditional metric is an absolute metric which reflects
  47 -- interactions of 2 terms in the corpus.
  48
  49 -- Filtered with MiniMax.
  50
  51 measureConditional :: Matrix Int -> Matrix Double
  52 measureConditional m =  run $ x $ map fromIntegral $ use m
  53   where
  54     x :: Acc (Matrix Double) -> Acc (Matrix Double)
  55     x mat = matMiniMax $ matProba r mat
  56
  57     r :: Dim
  58     r = dim m
  59
  60
  61 -- | To filter the nodes
  62 -- The conditional metric P(i|j) of 2 terms @i@ and @j@, also called
  63 -- "confidence" , is the maximum probability between @i@ and @j@ to see
  64 -- @i@ in the same context of @j@ knowing @j@.
  65 --
  66 -- If N(i) (resp. N(j)) is the number of occurrences of @i@ (resp. @j@)
  67 -- in the corpus and _[n_{ij}\] the number of its occurrences we get:
  68 --
  69 -- \[P_c=max(\frac{n_i}{n_{ij}},\frac{n_j}{n_{ij}} )\]
  70 conditional' :: Matrix Int -> (Matrix GenericityInclusion, Matrix SpecificityExclusion)
  71 conditional' m = ( run $ ie $ map fromIntegral $ use m
  72                  , run $ sg $ map fromIntegral $ use m
  73                  )
  74   where
  75     x :: Acc (Matrix Double) -> Acc (Matrix Double)
  76     x mat = (matProba r mat)
  77
  78     xs :: Acc (Matrix Double) -> Acc (Matrix Double)
  79     xs mat = let mat' = x mat in zipWith (-) (matSumLin r mat') mat'
  80     ys :: Acc (Matrix Double) -> Acc (Matrix Double)
  81     ys mat = let mat' = x mat in zipWith (-) (matSumCol r mat') mat'
  82
  83
  84     ie :: Acc (Matrix Double) -> Acc (Matrix Double)
  85     ie mat = map (\x' -> x' / (2*(n-1))) $ zipWith (+) (xs mat) (ys mat)
  86     sg :: Acc (Matrix Double) -> Acc (Matrix Double)
  87     sg mat = map (\x' -> x' / (2*(n-1))) $ zipWith (-) (xs mat) (ys mat)
  88
  89     r :: Dim
  90     r = dim m
  91
  92     n :: Exp Double
  93     n = P.fromIntegral r
  94