src/Gargantext/Viz/AdaptativePhylo.hs

   1 {-|
   2 Module      : Gargantext.Viz.AdaptativePhylo
   3 Description : Phylomemy definitions and types.
   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 export 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 References:
  19 Chavalarias, D., Cointet, J.-P., 2013. Phylomemetic patterns
  20 in science evolution — the rise and fall of scientific fields. PloS
  21 one 8, e54847.
  22 -}
  23
  24 {-# LANGUAGE DeriveAnyClass   #-}
  25 {-# LANGUAGE TemplateHaskell   #-}
  26
  27 module Gargantext.Viz.AdaptativePhylo where
  28
  29 import Data.Aeson
  30 import Data.Aeson.TH (deriveJSON)
  31 import Data.Text   (Text, pack)
  32 import Data.Vector (Vector)
  33 import Data.Map (Map)
  34 import Data.Set (Set)
  35
  36 import Gargantext.Core.Utils.Prefix (unPrefix)
  37 import Gargantext.Prelude
  38 import Gargantext.Text.Context (TermList)
  39
  40 import GHC.Generics
  41 import GHC.IO (FilePath)
  42 import Control.DeepSeq (NFData)
  43 import Control.Lens (makeLenses)
  44
  45 import qualified Data.Text.Lazy as TextLazy
  46
  47
  48 ----------------
  49 -- | Config | --
  50 ----------------
  51
  52
  53 data CorpusParser =
  54       Wos {_wos_limit :: Int}
  55     | Csv {_csv_limit :: Int}
  56     deriving (Show,Generic,Eq)
  57
  58 data SeaElevation =
  59       Constante
  60       { _cons_start :: Double
  61       , _cons_step  :: Double }
  62     | Adaptative
  63       { _adap_granularity :: Double }
  64     deriving (Show,Generic,Eq)
  65
  66 data Proximity =
  67       WeightedLogJaccard
  68       { _wlj_sensibility   :: Double
  69 {-
  70       -- , _wlj_thresholdInit :: Double
  71       -- , _wlj_thresholdStep :: Double
  72       -- | max height for sea level in temporal matching
  73       -- , _wlj_elevation     :: Double
  74 -}
  75       }
  76     | Hamming
  77     deriving (Show,Generic,Eq)
  78
  79
  80 data SynchronyScope = SingleBranch | SiblingBranches | AllBranches deriving (Show,Generic,Eq)
  81
  82 data SynchronyStrategy = MergeRegularGroups | MergeAllGroups deriving (Show,Generic,Eq)
  83
  84 data Synchrony =
  85       ByProximityThreshold
  86       { _bpt_threshold :: Double
  87       , _bpt_sensibility :: Double
  88       , _bpt_scope :: SynchronyScope
  89       , _bpt_strategy :: SynchronyStrategy }
  90     | ByProximityDistribution
  91       { _bpd_sensibility :: Double
  92       , _bpd_strategy :: SynchronyStrategy }
  93     deriving (Show,Generic,Eq)
  94
  95
  96 data TimeUnit =
  97       Year
  98       { _year_period :: Int
  99       , _year_step   :: Int
 100       , _year_matchingFrame :: Int }
 101       deriving (Show,Generic,Eq)
 102
 103
 104 data Clique =
 105       Fis
 106       { _fis_support :: Int
 107       , _fis_size    :: Int }
 108     | MaxClique
 109       { _mcl_size :: Int }
 110       deriving (Show,Generic,Eq)
 111
 112
 113 data Quality =
 114      Quality { _qua_granularity :: Double
 115              , _qua_minBranch   :: Int }
 116       deriving (Show,Generic,Eq)
 117
 118
 119 data Config =
 120      Config { corpusPath     :: FilePath
 121             , listPath       :: FilePath
 122             , outputPath     :: FilePath
 123             , corpusParser   :: CorpusParser
 124             , phyloName      :: Text
 125             , phyloLevel     :: Int
 126             , phyloProximity :: Proximity
 127             , seaElevation   :: SeaElevation
 128             , phyloSynchrony :: Synchrony
 129             , phyloQuality   :: Quality
 130             , timeUnit       :: TimeUnit
 131             , clique         :: Clique
 132             , exportLabel    :: [PhyloLabel]
 133             , exportSort     :: Sort
 134             , exportFilter   :: [Filter]
 135             } deriving (Show,Generic,Eq)
 136
 137
 138 defaultConfig :: Config
 139 defaultConfig =
 140      Config { corpusPath     = ""
 141             , listPath       = ""
 142             , outputPath     = ""
 143             , corpusParser   = Csv 1000
 144             , phyloName      = pack "Default Phylo"
 145             , phyloLevel     = 2
 146             , phyloProximity = WeightedLogJaccard 10
 147             , seaElevation   = Constante 0 0.1
 148             , phyloSynchrony = ByProximityThreshold 0.5 10 SiblingBranches MergeAllGroups
 149             , phyloQuality   = Quality 0.6 1
 150             , timeUnit       = Year 3 1 5
 151             , clique         = Fis 1 5
 152             , exportLabel    = [BranchLabel MostInclusive 2, GroupLabel MostEmergentInclusive 2]
 153             , exportSort     = ByHierarchy
 154             , exportFilter   = [ByBranchSize 2]
 155             }
 156
 157 instance FromJSON Config
 158 instance ToJSON Config
 159 instance FromJSON CorpusParser
 160 instance ToJSON CorpusParser
 161 instance FromJSON Proximity
 162 instance ToJSON Proximity
 163 instance FromJSON SeaElevation
 164 instance ToJSON SeaElevation
 165 instance FromJSON TimeUnit
 166 instance ToJSON TimeUnit
 167 instance FromJSON Clique
 168 instance ToJSON Clique
 169 instance FromJSON PhyloLabel
 170 instance ToJSON PhyloLabel
 171 instance FromJSON Tagger
 172 instance ToJSON Tagger
 173 instance FromJSON Sort
 174 instance ToJSON Sort
 175 instance FromJSON Order
 176 instance ToJSON Order
 177 instance FromJSON Filter
 178 instance ToJSON Filter
 179 instance FromJSON SynchronyScope
 180 instance ToJSON SynchronyScope
 181 instance FromJSON SynchronyStrategy
 182 instance ToJSON SynchronyStrategy
 183 instance FromJSON Synchrony
 184 instance ToJSON Synchrony
 185 instance FromJSON Quality
 186 instance ToJSON Quality
 187
 188
 189 -- | Software parameters
 190 data Software =
 191      Software { _software_name    :: Text
 192               , _software_version :: Text
 193      } deriving (Generic, Show, Eq)
 194
 195 defaultSoftware :: Software
 196 defaultSoftware =
 197       Software { _software_name    = pack "Gargantext"
 198                , _software_version = pack "v4" }
 199
 200
 201 -- | Global parameters of a Phylo
 202 data PhyloParam =
 203      PhyloParam { _phyloParam_version  :: Text
 204                 , _phyloParam_software :: Software
 205                 , _phyloParam_config   :: Config
 206      } deriving (Generic, Show, Eq)
 207
 208 defaultPhyloParam :: PhyloParam
 209 defaultPhyloParam =
 210       PhyloParam { _phyloParam_version  = pack "v2.adaptative"
 211                  , _phyloParam_software = defaultSoftware
 212                  , _phyloParam_config   = defaultConfig }
 213
 214
 215 ------------------
 216 -- | Document | --
 217 ------------------
 218
 219
 220 -- | Date : a simple Integer
 221 type Date = Int
 222
 223 -- | Ngrams : a contiguous sequence of n terms
 224 type Ngrams = Text
 225
 226 -- | Document : a piece of Text linked to a Date
 227 data Document = Document
 228       { date :: Date
 229       , text :: [Ngrams]
 230       } deriving (Eq,Show,Generic,NFData)
 231
 232
 233 --------------------
 234 -- | Foundation | --
 235 --------------------
 236
 237
 238 -- | The Foundations of a Phylo created from a given TermList
 239 data PhyloFoundations = PhyloFoundations
 240       { _foundations_roots   :: !(Vector Ngrams)
 241       , _foundations_mapList :: TermList
 242       } deriving (Generic, Show, Eq)
 243
 244
 245 ---------------------------
 246 -- | Coocurency Matrix | --
 247 ---------------------------
 248
 249
 250 -- | Cooc : a coocurency matrix between two ngrams
 251 type Cooc =  Map (Int,Int) Double
 252
 253
 254 -------------------
 255 -- | Phylomemy | --
 256 -------------------
 257
 258
 259 -- | Phylo datatype of a phylomemy
 260 --  foundations : the foundations of the phylo
 261 --  timeCooc : a Map of coocurency by minimal unit of time (ex: by year)
 262 --  timeDocs : a Map with the numbers of docs by minimal unit of time (ex: by year)
 263 --  param : the parameters of the phylomemy (with the user's configuration)
 264 --  periods : the temporal steps of a phylomemy
 265 data Phylo =
 266      Phylo { _phylo_foundations :: PhyloFoundations
 267            , _phylo_timeCooc    :: !(Map Date Cooc)
 268            , _phylo_timeDocs    :: !(Map Date Double)
 269            , _phylo_termFreq    :: !(Map Int Double)
 270            , _phylo_groupsProxi :: !(Map (PhyloGroupId,PhyloGroupId) Double)
 271            , _phylo_param       :: PhyloParam
 272            , _phylo_periods     :: Map PhyloPeriodId PhyloPeriod
 273            }
 274            deriving (Generic, Show, Eq)
 275
 276
 277 -- | PhyloPeriodId : the id of a given period
 278 type PhyloPeriodId = (Date,Date)
 279
 280 -- | PhyloPeriod : steps of a phylomemy on a temporal axis
 281 --  id: tuple (start date, end date) of the temporal step of the phylomemy
 282 --  levels: levels of granularity
 283 data PhyloPeriod =
 284      PhyloPeriod { _phylo_periodPeriod :: (Date,Date)
 285                  , _phylo_periodLevels :: Map PhyloLevelId PhyloLevel
 286                  } deriving (Generic, Show, Eq)
 287
 288
 289 -- | Level : a level of clustering
 290 type Level = Int
 291
 292 -- | PhyloLevelId : the id of a level of clustering in a given period
 293 type PhyloLevelId  = (PhyloPeriodId,Level)
 294
 295 -- | PhyloLevel : levels of phylomemy on a synchronic axis
 296 -- Levels description:
 297 -- Level 0: The foundations and the base of the phylo
 298 -- Level 1: First level of clustering (the Fis)
 299 -- Level [2..N]: Nth level of synchronic clustering (cluster of Fis)
 300 data PhyloLevel =
 301      PhyloLevel { _phylo_levelPeriod :: (Date,Date)
 302                 , _phylo_levelLevel  :: Level
 303                 , _phylo_levelGroups :: Map PhyloGroupId PhyloGroup
 304                 }
 305                 deriving (Generic, Show, Eq)
 306
 307
 308 type PhyloGroupId  = (PhyloLevelId, Int)
 309
 310 -- | BranchId : (a level, a sequence of branch index)
 311 -- the sequence is a path of heritage from the most to the less specific branch
 312 type PhyloBranchId = (Level, [Int])
 313
 314 -- | PhyloGroup : group of ngrams at each level and period
 315 data PhyloGroup =
 316       PhyloGroup { _phylo_groupPeriod   :: (Date,Date)
 317                  , _phylo_groupLevel    :: Level
 318                  , _phylo_groupIndex    :: Int
 319                  , _phylo_groupLabel    :: Text
 320                  , _phylo_groupSupport  :: Support
 321                  , _phylo_groupNgrams   :: [Int]
 322                  , _phylo_groupCooc     :: !(Cooc)
 323                  , _phylo_groupBranchId :: PhyloBranchId
 324                  , _phylo_groupMeta     :: Map Text [Double]
 325                  , _phylo_groupLevelParents  :: [Pointer]
 326                  , _phylo_groupLevelChilds   :: [Pointer]
 327                  , _phylo_groupPeriodParents :: [Pointer]
 328                  , _phylo_groupPeriodChilds  :: [Pointer]
 329                  }
 330                  deriving (Generic, Show, Eq, NFData)
 331
 332 -- | Weight : A generic mesure that can be associated with an Id
 333 type Weight = Double
 334
 335 -- | Pointer : A weighted pointer to a given PhyloGroup
 336 type Pointer = (PhyloGroupId, Weight)
 337
 338 data Filiation = ToParents | ToChilds deriving (Generic, Show)
 339 data PointerType = TemporalPointer | LevelPointer deriving (Generic, Show)
 340
 341
 342 ----------------------
 343 -- | Phylo Clique | --
 344 ----------------------
 345
 346 -- | Support : Number of Documents where a Clique occurs
 347 type Support  = Int
 348
 349 data PhyloClique = PhyloClique
 350   { _phyloClique_nodes   :: Set Ngrams
 351   , _phyloClique_support :: Support
 352   , _phyloClique_period  :: (Date,Date)
 353   } deriving (Generic,NFData,Show,Eq)
 354
 355
 356 ----------------
 357 -- | Export | --
 358 ----------------
 359
 360 type DotId = TextLazy.Text
 361
 362 data EdgeType = GroupToGroup | BranchToGroup | BranchToBranch | PeriodToPeriod deriving (Show,Generic,Eq)
 363
 364 data Filter = ByBranchSize { _branch_size :: Double } deriving (Show,Generic,Eq)
 365
 366 data Order = Asc | Desc deriving (Show,Generic,Eq)
 367
 368 data Sort = ByBirthDate { _sort_order :: Order } | ByHierarchy deriving (Show,Generic,Eq)
 369
 370 data Tagger = MostInclusive | MostEmergentInclusive deriving (Show,Generic,Eq)
 371
 372 data PhyloLabel =
 373       BranchLabel
 374       { _branch_labelTagger :: Tagger
 375       , _branch_labelSize   :: Int }
 376     | GroupLabel
 377       { _group_labelTagger  :: Tagger
 378       , _group_labelSize    :: Int }
 379     deriving (Show,Generic,Eq)
 380
 381 data PhyloBranch =
 382       PhyloBranch
 383       { _branch_id :: PhyloBranchId
 384       , _branch_canonId  :: [Int]
 385       , _branch_seaLevel :: [Double]
 386       , _branch_x        :: Double
 387       , _branch_y        :: Double
 388       , _branch_w        :: Double
 389       , _branch_t        :: Double
 390       , _branch_label    :: Text
 391       , _branch_meta     :: Map Text [Double]
 392       } deriving (Generic, Show, Eq)
 393
 394 data PhyloExport =
 395       PhyloExport
 396       { _export_groups   :: [PhyloGroup]
 397       , _export_branches :: [PhyloBranch]
 398       } deriving (Generic, Show)
 399
 400 ----------------
 401 -- | Lenses | --
 402 ----------------
 403
 404 makeLenses ''Config
 405 makeLenses ''Proximity
 406 makeLenses ''SeaElevation
 407 makeLenses ''Quality
 408 makeLenses ''Clique
 409 makeLenses ''PhyloLabel
 410 makeLenses ''TimeUnit
 411 makeLenses ''PhyloFoundations
 412 makeLenses ''PhyloClique
 413 makeLenses ''Phylo
 414 makeLenses ''PhyloPeriod
 415 makeLenses ''PhyloLevel
 416 makeLenses ''PhyloGroup
 417 makeLenses ''PhyloParam
 418 makeLenses ''PhyloExport
 419 makeLenses ''PhyloBranch
 420
 421 ------------------------
 422 -- | JSON instances | --
 423 ------------------------
 424
 425
 426 $(deriveJSON (unPrefix "_foundations_"  ) ''PhyloFoundations)