src/Gargantext/Viz/Phylo.hs

   1 {-|
   2 Module      : Gargantext.Viz.Phylo
   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
  25 {-# LANGUAGE DeriveGeneric     #-}
  26 {-# LANGUAGE NoImplicitPrelude #-}
  27 {-# LANGUAGE TemplateHaskell   #-}
  28 {-# LANGUAGE MultiParamTypeClasses #-}
  29
  30 module Gargantext.Viz.Phylo where
  31
  32 import Prelude (Bounded)
  33 import Control.Lens (makeLenses)
  34 import Data.Aeson.TH (deriveJSON,defaultOptions)
  35 import Data.Maybe   (Maybe)
  36 import Data.Text    (Text)
  37 import Data.Set     (Set)
  38 import Data.Map     (Map)
  39 import Data.Vector  (Vector)
  40 --import Data.Time.Clock.POSIX  (POSIXTime)
  41 import GHC.Generics (Generic)
  42 --import Gargantext.Database.Schema.Ngrams (NgramsId)
  43 import Gargantext.Core.Utils.Prefix (unPrefix)
  44 import Gargantext.Prelude
  45
  46 --------------------
  47 -- | PhyloParam | --
  48 --------------------
  49
  50
  51 -- | Global parameters of a Phylo
  52 data PhyloParam =
  53      PhyloParam { _phyloParam_version  :: Text -- Double ?
  54                 , _phyloParam_software :: Software
  55                 , _phyloParam_query    :: PhyloQueryBuild
  56      } deriving (Generic, Show, Eq)
  57
  58
  59 -- | Software parameters
  60 data Software =
  61      Software { _software_name    :: Text
  62               , _software_version :: Text
  63      } deriving (Generic, Show, Eq)
  64
  65
  66 ---------------
  67 -- | Phylo | --
  68 ---------------
  69
  70
  71 -- | Phylo datatype of a phylomemy
  72 -- Duration    : time Segment of the whole Phylo
  73 -- Foundations : vector of all the Ngrams contained in a Phylo (build from a list of actants)
  74 -- Periods     : list of all the periods of a Phylo
  75 data Phylo =
  76      Phylo { _phylo_duration    :: (Start, End)
  77            , _phylo_foundations :: Vector Ngrams
  78            , _phylo_foundationsPeaks :: PhyloPeaks
  79            , _phylo_periods     :: [PhyloPeriod]
  80            , _phylo_param       :: PhyloParam
  81            }
  82            deriving (Generic, Show, Eq)
  83
  84 -- | The PhyloPeaks describe the aggregation of some foundations Ngrams behind a list of Ngrams trees (ie: a forest)
  85 -- PeaksLabels are the root labels of each Ngrams trees
  86 data PhyloPeaks =
  87       PhyloPeaks { _phylo_peaksLabels :: Vector Ngrams
  88                  , _phylo_peaksForest :: [Tree Ngrams]
  89                  }
  90                  deriving (Generic, Show, Eq)
  91
  92 -- | A Tree of Ngrams where each node is a label
  93 data Tree a = Empty | Node a [Tree a] deriving (Generic, Show, Eq)
  94
  95
  96 -- | Date : a simple Integer
  97 type Date = Int
  98
  99 -- | UTCTime in seconds since UNIX epoch
 100 -- type Start   = POSIXTime
 101 -- type End     = POSIXTime
 102 type Start   = Date
 103 type End     = Date
 104
 105
 106 ---------------------
 107 -- | PhyloPeriod | --
 108 ---------------------
 109
 110
 111 -- | PhyloStep : steps of phylomemy on temporal axis
 112 -- Period: tuple (start date, end date) of the step of the phylomemy
 113 -- Levels: levels of granularity
 114 data PhyloPeriod =
 115      PhyloPeriod { _phylo_periodId     :: PhyloPeriodId
 116                  , _phylo_periodLevels :: [PhyloLevel]
 117                  }
 118                  deriving (Generic, Show, Eq)
 119
 120
 121 --------------------
 122 -- | PhyloLevel | --
 123 --------------------
 124
 125
 126 -- | PhyloLevel : levels of phylomemy on level axis
 127 -- Levels description:
 128 -- Level -1: Ngram equals itself         (by identity) == _phylo_Ngrams
 129 -- Level  0: Group of synonyms           (by stems + by qualitative expert meaning)
 130 -- Level  1: First level of clustering
 131 -- Level  N: Nth   level of clustering
 132 data PhyloLevel =
 133      PhyloLevel { _phylo_levelId     :: PhyloLevelId
 134                 , _phylo_levelGroups :: [PhyloGroup]
 135                 }
 136                 deriving (Generic, Show, Eq)
 137
 138
 139 --------------------
 140 -- | PhyloGroup | --
 141 --------------------
 142
 143
 144 -- | PhyloGroup : group of ngrams at each level and step
 145 -- Label : maybe has a label as text
 146 -- Ngrams: set of terms that build the group
 147 -- Quality : map of measures (support, etc.) that depict some qualitative aspects of a phylo
 148 -- Period Parents|Childs: weighted link to Parents|Childs (Temporal Period   axis)
 149 -- Level  Parents|Childs: weighted link to Parents|Childs (Level Granularity axis)
 150 -- Pointers are directed link from Self to any PhyloGroup (/= Self ?)
 151 data PhyloGroup =
 152      PhyloGroup { _phylo_groupId            :: PhyloGroupId
 153                 , _phylo_groupLabel         :: Text
 154                 , _phylo_groupNgrams        :: [Int]
 155                 , _phylo_groupMeta          :: Map Text Double
 156                 , _phylo_groupCooc          :: Map (Int, Int) Double
 157                 , _phylo_groupBranchId      :: Maybe PhyloBranchId
 158
 159                 , _phylo_groupPeriodParents :: [Pointer]
 160                 , _phylo_groupPeriodChilds  :: [Pointer]
 161
 162                 , _phylo_groupLevelParents  :: [Pointer]
 163                 , _phylo_groupLevelChilds   :: [Pointer]
 164                 }
 165                 deriving (Generic, Show, Eq, Ord)
 166
 167
 168 -- | Level : A level of aggregation (-1 = Txt, 0 = Ngrams, 1 = Fis, [2..] = Cluster)
 169 type Level = Int
 170 -- | Index : A generic index of an element (PhyloGroup, PhyloBranch, etc) in a given List
 171 type Index = Int
 172
 173
 174 type PhyloPeriodId = (Start, End)
 175 type PhyloLevelId  = (PhyloPeriodId, Level)
 176 type PhyloGroupId  = (PhyloLevelId, Index)
 177 type PhyloBranchId = (Level, Index)
 178
 179
 180 -- | Weight : A generic mesure that can be associated with an Id
 181 type Weight = Double
 182 -- | Pointer : A weighted linked with a given PhyloGroup
 183 type Pointer = (PhyloGroupId, Weight)
 184 -- | Ngrams : a contiguous sequence of n terms
 185 type Ngrams = Text
 186
 187
 188 --------------------
 189 -- | Aggregates | --
 190 --------------------
 191
 192
 193 -- | Document : a piece of Text linked to a Date
 194 data Document = Document
 195       { date :: Date
 196       , text :: [Ngrams]
 197       } deriving (Show)
 198
 199 -- | Clique : Set of ngrams cooccurring in the same Document
 200 type Clique   = Set Ngrams
 201 -- | Support : Number of Documents where a Clique occurs
 202 type Support  = Int
 203 -- | Fis : Frequent Items Set (ie: the association between a Clique and a Support)
 204 data PhyloFis = PhyloFis
 205   { _phyloFis_clique  :: Clique
 206   , _phyloFis_support :: Support
 207   , _phyloFis_metrics :: Map (Int,Int) (Map Text [Double])
 208   } deriving (Show)
 209
 210 -- | A list of clustered PhyloGroup
 211 type PhyloCluster = [PhyloGroup]
 212
 213
 214 -- | A List of PhyloGroup in a Graph
 215 type GroupNodes = [PhyloGroup]
 216 -- | A List of weighted links between some PhyloGroups in a Graph
 217 type GroupEdges = [((PhyloGroup,PhyloGroup),Weight)]
 218 -- | The association as a Graph between a list of Nodes and a list of Edges
 219 type GroupGraph = (GroupNodes,GroupEdges)
 220
 221
 222 ---------------
 223 -- | Error | --
 224 ---------------
 225
 226
 227 data PhyloError = LevelDoesNotExist
 228                 | LevelUnassigned
 229           deriving (Show)
 230
 231
 232 -----------------
 233 -- | Cluster | --
 234 -----------------
 235
 236
 237 -- | Cluster constructors
 238 data Cluster = Fis FisParams
 239              | RelatedComponents RCParams
 240              | Louvain LouvainParams
 241         deriving (Generic, Show, Eq, Read)
 242
 243 -- | Parameters for Fis clustering
 244 data FisParams = FisParams
 245   { _fis_keepMinorFis :: Bool
 246   , _fis_minSupport   :: Support
 247   } deriving (Generic, Show, Eq, Read)
 248
 249 -- | Parameters for RelatedComponents clustering
 250 data RCParams = RCParams
 251   { _rc_proximity :: Proximity } deriving (Generic, Show, Eq, Read)
 252
 253 -- | Parameters for Louvain clustering
 254 data LouvainParams = LouvainParams
 255   { _louvain_proximity :: Proximity } deriving (Generic, Show, Eq, Read)
 256
 257
 258 -------------------
 259 -- | Proximity | --
 260 -------------------
 261
 262
 263 -- | Proximity constructors
 264 data Proximity = WeightedLogJaccard WLJParams
 265                | Hamming HammingParams
 266                | Filiation
 267           deriving (Generic, Show, Eq, Read)
 268
 269 -- | Parameters for WeightedLogJaccard proximity
 270 data WLJParams = WLJParams
 271   { _wlj_threshold   :: Double
 272   , _wlj_sensibility :: Double
 273   } deriving (Generic, Show, Eq, Read)
 274
 275 -- | Parameters for Hamming proximity
 276 data HammingParams = HammingParams
 277   { _hamming_threshold :: Double } deriving (Generic, Show, Eq, Read)
 278
 279
 280 ----------------
 281 -- | Filter | --
 282 ----------------
 283
 284
 285 -- | Filter constructors
 286 data Filter = SmallBranch SBParams deriving (Generic, Show, Eq)
 287
 288 -- | Parameters for SmallBranch filter
 289 data SBParams = SBParams
 290   { _sb_periodsInf :: Int
 291   , _sb_periodsSup :: Int
 292   , _sb_minNodes   :: Int } deriving (Generic, Show, Eq)
 293
 294
 295 ----------------
 296 -- | Metric | --
 297 ----------------
 298
 299
 300 -- | Metric constructors
 301 data Metric = BranchAge deriving (Generic, Show, Eq, Read)
 302
 303
 304 ----------------
 305 -- | Tagger | --
 306 ----------------
 307
 308
 309 -- | Tagger constructors
 310 data Tagger = BranchLabelFreq | GroupLabelCooc | GroupDynamics
 311   deriving (Generic, Show, Read)
 312
 313
 314 --------------
 315 -- | Sort | --
 316 --------------
 317
 318
 319 -- | Sort constructors
 320 data Sort  = ByBranchAge deriving (Generic, Show, Read, Enum, Bounded)
 321 data Order = Asc | Desc  deriving (Generic, Show, Read)
 322
 323
 324 --------------------
 325 -- | PhyloQuery | --
 326 --------------------
 327
 328
 329 -- | A Phyloquery describes a phylomemic reconstruction
 330 data PhyloQueryBuild = PhyloQueryBuild
 331     { _q_phyloTitle :: Text
 332     , _q_phyloDesc  :: Text
 333
 334     -- Grain and Steps for the PhyloPeriods
 335     , _q_periodGrain :: Int
 336     , _q_periodSteps :: Int
 337
 338     -- Clustering method for building the contextual unit of Phylo (ie: level 1)
 339     , _q_contextualUnit :: Cluster
 340     , _q_contextualUnitMetrics :: [Metric]
 341     , _q_contextualUnitFilters :: [Filter]
 342
 343     -- Inter-temporal matching method of the Phylo
 344     , _q_interTemporalMatching :: Proximity
 345
 346     -- Last level of reconstruction
 347     , _q_nthLevel   :: Level
 348     -- Clustering method used from level 1 to nthLevel
 349     , _q_nthCluster :: Cluster
 350     } deriving (Generic, Show, Eq)
 351
 352 -- | To choose the Phylo edge you want to export : --> <-- <--> <=>
 353 data Filiation = Ascendant | Descendant | Merge | Complete deriving (Generic, Show, Read)
 354 data EdgeType  = PeriodEdge | LevelEdge deriving (Generic, Show)
 355
 356 -------------------
 357 -- | PhyloView | --
 358 -------------------
 359
 360
 361 -- | A PhyloView is the output type of a Phylo
 362 data PhyloView = PhyloView
 363   { _pv_param       :: PhyloParam
 364   , _pv_title       :: Text
 365   , _pv_description :: Text
 366   , _pv_filiation   :: Filiation
 367   , _pv_metrics     :: Map Text [Double]
 368   , _pv_branches    :: [PhyloBranch]
 369   , _pv_nodes       :: [PhyloNode]
 370   , _pv_edges       :: [PhyloEdge]
 371   } deriving (Generic, Show)
 372
 373 -- | A phyloview is made of PhyloBranches, edges and nodes
 374 data PhyloBranch = PhyloBranch
 375   { _pb_id      :: PhyloBranchId
 376   , _pb_label   :: Text
 377   , _pb_metrics :: Map Text [Double]
 378   } deriving (Generic, Show)
 379
 380 data PhyloEdge = PhyloEdge
 381   { _pe_source :: PhyloGroupId
 382   , _pe_target :: PhyloGroupId
 383   , _pe_type   :: EdgeType
 384   , _pe_weight :: Weight
 385   } deriving (Generic, Show)
 386
 387 data PhyloNode = PhyloNode
 388   { _pn_id        :: PhyloGroupId
 389   , _pn_bid  :: Maybe PhyloBranchId
 390   , _pn_label     :: Text
 391   , _pn_idx :: [Int]
 392   , _pn_ngrams    :: Maybe [Ngrams]
 393   , _pn_metrics      :: Map Text [Double]
 394   , _pn_parents :: Maybe [PhyloGroupId]
 395   , _pn_childs  :: [PhyloNode]
 396   } deriving (Generic, Show)
 397
 398 ------------------------
 399 -- | PhyloQueryView | --
 400 ------------------------
 401
 402
 403 data DisplayMode = Flat | Nested
 404   deriving (Generic, Show, Read)
 405
 406 -- | A PhyloQueryView describes a Phylo as an output view
 407 data PhyloQueryView = PhyloQueryView
 408   { _qv_lvl    :: Level
 409
 410   -- Does the PhyloGraph contain ascendant, descendant or a complete Filiation ? Complet redondant et merge (avec le max)
 411   , _qv_filiation :: Filiation
 412
 413   -- Does the PhyloGraph contain some levelChilds ? How deep must it go ?
 414   , _qv_levelChilds      :: Bool
 415   , _qv_levelChildsDepth :: Level
 416
 417   -- Ordered lists of filters, taggers and metrics to be applied to the PhyloGraph
 418   -- Firstly the metrics, then the filters and the taggers
 419   , _qv_metrics :: [Metric]
 420   , _qv_filters :: [Filter]
 421   , _qv_taggers :: [Tagger]
 422
 423   -- An asc or desc sort to apply to the PhyloGraph
 424   , _qv_sort :: Maybe (Sort,Order)
 425
 426   -- A display mode to apply to the PhyloGraph, ie: [Node[Node,Edge],Edge] or [[Node,Node],[Edge,Edge]]
 427   , _qv_display :: DisplayMode
 428   , _qv_verbose :: Bool
 429   }
 430
 431
 432 ----------------
 433 -- | Lenses | --
 434 ----------------
 435
 436
 437 makeLenses ''PhyloParam
 438 makeLenses ''Software
 439 --
 440 makeLenses ''Phylo
 441 makeLenses ''PhyloPeaks
 442 makeLenses ''PhyloGroup
 443 makeLenses ''PhyloLevel
 444 makeLenses ''PhyloPeriod
 445 makeLenses ''PhyloFis
 446 --
 447 makeLenses ''Proximity
 448 makeLenses ''Cluster
 449 makeLenses ''Filter
 450 --
 451 makeLenses ''PhyloQueryBuild
 452 makeLenses ''PhyloQueryView
 453 --
 454 makeLenses ''PhyloView
 455 makeLenses ''PhyloBranch
 456 makeLenses ''PhyloNode
 457 makeLenses ''PhyloEdge
 458
 459
 460 ------------------------
 461 -- | JSON instances | --
 462 ------------------------
 463
 464
 465 $(deriveJSON (unPrefix "_phylo_"       ) ''Phylo       )
 466 $(deriveJSON (unPrefix "_phylo_peaks"  ) ''PhyloPeaks  )
 467 $(deriveJSON defaultOptions ''Tree  )
 468 $(deriveJSON (unPrefix "_phylo_period" ) ''PhyloPeriod )
 469 $(deriveJSON (unPrefix "_phylo_level"  ) ''PhyloLevel  )
 470 $(deriveJSON (unPrefix "_phylo_group"  ) ''PhyloGroup  )
 471 $(deriveJSON (unPrefix "_phyloFis_"    ) ''PhyloFis    )
 472 --
 473 $(deriveJSON (unPrefix "_software_"    ) ''Software    )
 474 $(deriveJSON (unPrefix "_phyloParam_"  ) ''PhyloParam  )
 475 --
 476 $(deriveJSON defaultOptions ''Filter    )
 477 $(deriveJSON defaultOptions ''Metric    )
 478 $(deriveJSON defaultOptions ''Cluster   )
 479 $(deriveJSON defaultOptions ''Proximity )
 480 --
 481 $(deriveJSON (unPrefix "_fis_" )     ''FisParams     )
 482 $(deriveJSON (unPrefix "_hamming_" ) ''HammingParams )
 483 $(deriveJSON (unPrefix "_louvain_" ) ''LouvainParams )
 484 $(deriveJSON (unPrefix "_rc_" )      ''RCParams      )
 485 $(deriveJSON (unPrefix "_wlj_" )     ''WLJParams     )
 486 $(deriveJSON (unPrefix "_sb_" )      ''SBParams      )
 487 --
 488 $(deriveJSON (unPrefix "_q_" )  ''PhyloQueryBuild  )
 489 $(deriveJSON (unPrefix "_pv_" ) ''PhyloView   )
 490 $(deriveJSON (unPrefix "_pb_" ) ''PhyloBranch )
 491 $(deriveJSON (unPrefix "_pe_" ) ''PhyloEdge   )
 492 $(deriveJSON (unPrefix "_pn_" ) ''PhyloNode   )
 493
 494 $(deriveJSON defaultOptions ''Filiation )
 495 $(deriveJSON defaultOptions ''EdgeType  )
 496
 497
 498 ----------------------------
 499 -- | TODO XML instances | --
 500 ----------------------------
 501