src/Gargantext/Types/Phylo.hs

   1 {-|
   2 Module      : Gargantext.Types.Phylo
   3 Description : Main Types for Phylomemy
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Specifications of Phylomemy format.
  11
  12 Phylomemy can be described as a Temporal Graph with different scale of
  13 granularity of group of ngrams (terms and multi-terms).
  14
  15 The main type is Phylo which is synonym of Phylomemy (only difference is
  16 the number of chars).
  17
  18 Phylomemy was first described in [REF].
  19 -}
  20
  21 {-# LANGUAGE DeriveGeneric        #-}
  22 {-# LANGUAGE TemplateHaskell      #-}
  23
  24 module Gargantext.Types.Phylo where
  25
  26 import Data.Aeson.TH (deriveJSON)
  27 import Data.Maybe   (Maybe)
  28 import Data.Text    (Text)
  29 import Data.Time.Clock.POSIX  (POSIXTime)
  30
  31 import GHC.Generics (Generic)
  32
  33 import Gargantext.Prelude
  34 import Gargantext.Utils.Prefix (unPrefix)
  35
  36 ------------------------------------------------------------------------
  37 -- | Phylo datatype descriptor of a phylomemy
  38 -- Duration : time Segment of the whole phylomemy in UTCTime format (start,end)
  39 -- Ngrams   : list of all (possible) terms contained in the phylomemy (with their id)
  40 -- Steps    : list of all steps to build the phylomemy
  41 data Phylo = Phylo { _phyloDuration :: (Start, End)
  42                    , _phyloNgrams   :: [Ngram]
  43                    , _phyloPeriods  :: [PhyloPeriod]
  44                    } deriving (Generic)
  45
  46 -- | UTCTime in seconds since UNIX epoch
  47 type Start   = POSIXTime
  48 type End     = POSIXTime
  49
  50 type Ngram   = (NgramId, Text)
  51 type NgramId = Int
  52
  53 -- | PhyloStep : steps of phylomemy on temporal axis
  54 -- Period: tuple (start date, end date) of the step of the phylomemy
  55 -- Levels: levels of granularity
  56 data PhyloPeriod = PhyloPeriod { _phyloPeriodId     :: PhyloPeriodId
  57                                , _phyloPeriodLevels :: [PhyloLevel]
  58                                } deriving (Generic)
  59
  60 type PhyloPeriodId = (Start, End)
  61
  62 -- | PhyloLevel : levels of phylomemy on level axis
  63 -- Levels description:
  64 -- Level -1: Ngram equals itself         (by identity) == _phyloNgrams
  65 -- Level  0: Group of synonyms           (by stems + by qualitative expert meaning)
  66 -- Level  1: First level of clustering
  67 -- Level  N: Nth   level of clustering
  68 data PhyloLevel = PhyloLevel { _phyloLevelId     :: PhyloLevelId
  69                              , _phyloLevelGroups :: [PhyloGroup]
  70                              } deriving (Generic)
  71
  72 type PhyloLevelId = (PhyloPeriodId, Int)
  73
  74 -- | PhyloGroup : group of ngrams at each level and step
  75 -- Label : maybe has a label as text
  76 -- Ngrams: set of terms that build the group
  77 -- Period Parents|Childs: weighted link to Parents|Childs (Temporal Period   axis)
  78 -- Level  Parents|Childs: weighted link to Parents|Childs (Level Granularity axis)
  79 data PhyloGroup = PhyloGroup { _phyloGroupId            :: PhyloGroupId
  80                              , _phyloGroupLabel         :: Maybe Text
  81                              , _phyloGroupNgrams        :: [NgramId]
  82
  83                              , _phyloGroupPeriodParents :: [Edge]
  84                              , _phyloGroupPeriodChilds  :: [Edge]
  85
  86                              , _phyloGroupLevelParents  :: [Edge]
  87                              , _phyloGroupLevelChilds   :: [Edge]
  88                              } deriving (Generic)
  89
  90 type PhyloGroupId = (PhyloLevelId, Int)
  91 type Edge         = (PhyloGroupId, Weight)
  92 type Weight       = Double
  93
  94 -- | JSON instances
  95 $(deriveJSON (unPrefix "_phylo"       ) ''Phylo       )
  96 $(deriveJSON (unPrefix "_phyloPeriod" ) ''PhyloPeriod )
  97 $(deriveJSON (unPrefix "_phyloLevel"  ) ''PhyloLevel  )
  98 $(deriveJSON (unPrefix "_phyloGroup"  ) ''PhyloGroup  )