src/Gargantext/Viz/Phylo/SynchronicClustering.hs

   1 {-|
   2 Module      : Gargantext.Viz.Phylo.SynchronicClustering
   3 Description : Module dedicated to the adaptative synchronic clustering of a Phylo.
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9 -}
  10
  11 {-# LANGUAGE NoImplicitPrelude #-}
  12 {-# LANGUAGE FlexibleContexts  #-}
  13 {-# LANGUAGE OverloadedStrings #-}
  14 {-# LANGUAGE MultiParamTypeClasses #-}
  15
  16 module Gargantext.Viz.Phylo.SynchronicClustering where
  17
  18 import Gargantext.Prelude
  19 import Gargantext.Viz.AdaptativePhylo
  20 import Gargantext.Viz.Phylo.PhyloTools
  21 import Gargantext.Viz.Phylo.TemporalMatching (weightedLogJaccard', filterDiago, reduceDiagos)
  22 import Gargantext.Viz.Phylo.PhyloExport (processDynamics)
  23
  24 import Data.List ((++), null, intersect, nub, concat, sort, sortOn, init, all, group, maximum, groupBy)
  25 import Data.Map (Map, fromList, fromListWith, foldlWithKey, (!), insert, empty, restrictKeys, elems, mapWithKey, member, singleton)
  26
  27 import Control.Lens hiding (Level)
  28 import Control.Parallel.Strategies (parList, rdeepseq, using)
  29 -- import Debug.Trace (trace)
  30
  31 import qualified Data.Map as Map
  32 import qualified Data.Set as Set
  33
  34
  35 -------------------------
  36 -- | New Level Maker | --
  37 -------------------------
  38
  39 mergeBranchIds :: [[Int]] -> [Int]
  40 mergeBranchIds ids = (head' "mergeBranchIds" . sort . mostFreq) ids
  41   where
  42     -- | 2) find the most Up Left ids in the hierarchy of similarity
  43     -- mostUpLeft :: [[Int]] -> [[Int]]
  44     -- mostUpLeft ids' =
  45     --      let groupIds = (map (\gIds -> (length $ head' "gIds" gIds, head' "gIds" gIds)) . groupBy (\id id' -> length id == length id') . sortOn length) ids'
  46     --          inf = (fst . minimum) groupIds
  47     --       in map snd $ filter (\gIds -> fst gIds == inf) groupIds
  48     -- | 1) find the most frequent ids
  49     mostFreq :: [[Int]] -> [[Int]]
  50     mostFreq ids' =
  51       let groupIds = (map (\gIds -> (length gIds, head' "gIds" gIds)) . group . sort) ids'
  52           sup = (fst . maximum) groupIds
  53        in map snd $ filter (\gIds -> fst gIds == sup) groupIds
  54
  55
  56 groupsToBranches' :: Map PhyloGroupId PhyloGroup -> [[PhyloGroup]]
  57 groupsToBranches' groups =
  58     -- | run the related component algorithm
  59     let egos  = map (\g -> [getGroupId g]
  60                         ++ (map fst $ g ^. phylo_groupPeriodParents)
  61                         ++ (map fst $ g ^. phylo_groupPeriodChilds) ) $ elems groups
  62         graph = relatedComponents egos
  63     -- | update each group's branch id
  64     in map (\ids ->
  65         let groups' = elems $ restrictKeys groups (Set.fromList ids)
  66             bId = mergeBranchIds $ map (\g -> snd $ g ^. phylo_groupBranchId) groups'
  67          in map (\g -> g & phylo_groupBranchId %~ (\(lvl,_) -> (lvl + 1,bId))) groups') graph
  68
  69 getLastThr :: [PhyloGroup] -> Double
  70 getLastThr childs = maximum $ concat $ map (\g -> (g ^. phylo_groupMeta) ! "thr") childs
  71
  72 mergeGroups :: [Cooc] -> PhyloGroupId -> Map PhyloGroupId PhyloGroupId -> [PhyloGroup] -> PhyloGroup
  73 mergeGroups coocs id mapIds childs =
  74     let ngrams = (sort . nub . concat) $ map _phylo_groupNgrams childs
  75     in PhyloGroup (fst $ fst id) (snd $ fst id) (snd id)  ""
  76                   (sum $ map _phylo_groupSupport childs)  ngrams
  77                   (ngramsToCooc ngrams coocs)
  78                   ((snd $ fst id),(mergeBranchIds $ map (\g -> snd $ g ^. phylo_groupBranchId) childs))
  79                   (singleton "thr" [getLastThr childs]) [] (map (\g -> (getGroupId g, 1)) childs)
  80                   (updatePointers $ concat $ map _phylo_groupPeriodParents childs)
  81                   (updatePointers $ concat $ map _phylo_groupPeriodChilds  childs)
  82     where
  83         updatePointers :: [Pointer] -> [Pointer]
  84         updatePointers pointers = map (\(pId,w) -> (mapIds ! pId,w)) pointers
  85
  86
  87 addPhyloLevel :: Level -> Phylo -> Phylo
  88 addPhyloLevel lvl phylo =
  89   over ( phylo_periods .  traverse )
  90        (\phyloPrd -> phyloPrd & phylo_periodLevels
  91                         %~ (insert (phyloPrd ^. phylo_periodPeriod, lvl) (PhyloLevel (phyloPrd ^. phylo_periodPeriod) lvl empty))) phylo
  92
  93
  94 toNextLevel' :: Phylo -> [PhyloGroup] -> Phylo
  95 toNextLevel' phylo groups =
  96     let curLvl = getLastLevel phylo
  97         oldGroups = fromList $ map (\g -> (getGroupId g, getLevelParentId g)) groups
  98         newGroups = concat $ groupsToBranches'
  99                   $ fromList $ map (\g -> (getGroupId g, g))
 100                   $ foldlWithKey (\acc id groups' ->
 101                         -- | 4) create the parent group
 102                         let parent = mergeGroups (elems $ restrictKeys (phylo ^. phylo_timeCooc) $ periodsToYears [(fst . fst) id]) id oldGroups groups'
 103                         in  acc ++ [parent]) []
 104                   -- | 3) group the current groups by parentId
 105                   $ fromListWith (++) $ map (\g -> (getLevelParentId g, [g])) groups
 106
 107         newPeriods = fromListWith (++) $ map (\g -> (g ^. phylo_groupPeriod, [g])) newGroups
 108     in  traceSynchronyEnd
 109       $ over ( phylo_periods . traverse . phylo_periodLevels . traverse
 110              -- | 6) update each period at curLvl + 1
 111              . filtered (\phyloLvl -> phyloLvl ^. phylo_levelLevel == (curLvl + 1)))
 112              -- | 7) by adding the parents
 113              (\phyloLvl ->
 114                 if member (phyloLvl ^. phylo_levelPeriod) newPeriods
 115                     then phyloLvl & phylo_levelGroups
 116                             .~ fromList (map (\g -> (getGroupId g, g)) $ newPeriods ! (phyloLvl ^. phylo_levelPeriod))
 117                     else phyloLvl)
 118       -- | 2) add the curLvl + 1 phyloLevel to the phylo
 119       $ addPhyloLevel (curLvl + 1)
 120       -- | 1) update the current groups (with level parent pointers) in the phylo
 121       $ updatePhyloGroups curLvl (fromList $ map (\g -> (getGroupId g, g)) groups) phylo
 122
 123 --------------------
 124 -- | Clustering | --
 125 --------------------
 126
 127 toPairs :: SynchronyStrategy -> [PhyloGroup] -> [(PhyloGroup,PhyloGroup)]
 128 toPairs strategy groups = case strategy of
 129   MergeRegularGroups -> pairs
 130                       $ filter (\g -> all (== 3) $ (g ^. phylo_groupMeta) ! "dynamics") groups
 131   MergeAllGroups -> pairs groups
 132   where
 133     pairs :: [PhyloGroup] -> [(PhyloGroup,PhyloGroup)]
 134     pairs gs = filter (\(g,g') -> (not . null) $ intersect (g ^. phylo_groupNgrams) (g' ^. phylo_groupNgrams)) (listToCombi' gs)
 135
 136
 137 toDiamonds :: [PhyloGroup] -> [[PhyloGroup]]
 138 toDiamonds groups = foldl' (\acc groups' ->
 139                         acc ++ ( elems
 140                                $ Map.filter (\v -> length v > 1)
 141                                $ fromListWith (++)
 142                                $ foldl' (\acc' g ->
 143                                     acc' ++ (map (\(id,_) -> (id,[g]) ) $ g ^. phylo_groupPeriodChilds)) [] groups')) []
 144                   $ elems
 145                   $ Map.filter (\v -> length v > 1)
 146                   $ fromListWith (++)
 147                   $ foldl' (\acc g -> acc ++ (map (\(id,_) -> (id,[g]) ) $ g ^. phylo_groupPeriodParents)  ) [] groups
 148
 149
 150 groupsToEdges :: Proximity -> Synchrony -> Double -> Map Int Double -> [PhyloGroup] -> [((PhyloGroup,PhyloGroup),Double)]
 151 groupsToEdges prox sync nbDocs diago groups =
 152     case sync of
 153         ByProximityThreshold  thr sens _ strat ->
 154             filter (\(_,w) -> w >= thr)
 155           $ toEdges sens
 156           $ toPairs strat groups
 157         ByProximityDistribution sens strat ->
 158             let diamonds = sortOn snd
 159                          $ toEdges sens $ concat
 160                          $ map (\gs -> toPairs strat gs) $ toDiamonds groups
 161              in take (div (length diamonds) 2) diamonds
 162     where
 163         toEdges :: Double -> [(PhyloGroup,PhyloGroup)] -> [((PhyloGroup,PhyloGroup),Double)]
 164         toEdges sens edges =
 165             case prox of
 166                 WeightedLogJaccard _ _ _ -> map (\(g,g') ->
 167                                                  ((g,g'), weightedLogJaccard' sens nbDocs diago
 168                                                               (g ^. phylo_groupNgrams) (g' ^. phylo_groupNgrams))) edges
 169                 _ -> undefined
 170
 171
 172
 173 toRelatedComponents :: [PhyloGroup] -> [((PhyloGroup,PhyloGroup),Double)] -> [[PhyloGroup]]
 174 toRelatedComponents nodes edges =
 175   let ref = fromList $ map (\g -> (getGroupId g, g)) nodes
 176       clusters = relatedComponents $ ((map (\((g,g'),_) -> [getGroupId g, getGroupId g']) edges) ++ (map (\g -> [getGroupId g]) nodes))
 177    in map (\cluster -> map (\gId -> ref ! gId) cluster) clusters
 178
 179 toParentId :: PhyloGroup -> PhyloGroupId
 180 toParentId child = ((child ^. phylo_groupPeriod, child ^. phylo_groupLevel + 1), child ^. phylo_groupIndex)
 181
 182
 183 reduceGroups :: Proximity -> Synchrony -> Map Date Double -> Map Date Cooc -> [PhyloGroup] -> [PhyloGroup]
 184 reduceGroups prox sync docs diagos branch =
 185     -- | 1) reduce a branch as a set of periods & groups
 186     let periods = fromListWith (++)
 187                  $ map (\g -> (g ^. phylo_groupPeriod,[g])) branch
 188     in  (concat . concat . elems)
 189       $ mapWithKey (\prd groups ->
 190             -- | 2) for each period, transform the groups as a proximity graph filtered by a threshold
 191             let diago = reduceDiagos $ filterDiago diagos [prd]
 192                 edges = groupsToEdges prox sync ((sum . elems) $ restrictKeys docs $ periodsToYears [prd]) diago groups
 193              in map (\comp ->
 194                     -- | 4) add to each groups their futur level parent group
 195                     let parentId = toParentId (head' "parentId" comp)
 196                     in  map (\g -> g & phylo_groupLevelParents %~ (++ [(parentId,1)]) ) comp )
 197                 -- |3) reduce the graph a a set of related components
 198               $ toRelatedComponents groups edges) periods
 199
 200
 201 getGroupRealBId :: Double -> PhyloGroup -> [Int]
 202 getGroupRealBId step g =
 203   let nb = round(getGroupThr g / step) + 2
 204    in take nb (snd $ g ^. phylo_groupBranchId)
 205
 206
 207
 208 adjustClustering :: Synchrony -> Double -> [[PhyloGroup]] -> [[PhyloGroup]]
 209 adjustClustering sync step branches = case sync of
 210   ByProximityThreshold _ _ scope _ ->
 211     case scope of
 212       SingleBranch -> branches
 213       SiblingBranches -> groupBy (\g g' -> (init $ getGroupRealBId step g) == (init $ getGroupRealBId step g'))
 214                        $ sortOn _phylo_groupBranchId $ concat branches
 215       -- SiblingBranches -> elems $ fromListWith (++) $ map (\b -> ((init . snd . _phylo_groupBranchId) $ head' "adjustClustering" b,b)) branches
 216       AllBranches -> [concat branches]
 217   ByProximityDistribution _ _ -> branches
 218
 219
 220
 221 synchronicClustering :: Phylo -> Phylo
 222 synchronicClustering phylo =
 223     let prox = phyloProximity $ getConfig phylo
 224         sync = phyloSynchrony $ getConfig phylo
 225         docs = phylo ^. phylo_timeDocs
 226         diagos = map coocToDiago $ phylo ^. phylo_timeCooc
 227         newBranches  = map (\branch -> reduceGroups prox sync docs diagos branch)
 228                      $ map processDynamics
 229                      $ adjustClustering sync (getPhyloThresholdStep phylo)
 230                      $ phyloToLastBranches
 231                      $ traceSynchronyStart phylo
 232         newBranches' = newBranches `using` parList rdeepseq
 233      in toNextLevel' phylo $ concat newBranches'
 234
 235
 236 ----------------
 237 -- | probes | --
 238 ----------------
 239
 240 -- synchronicDistance :: Phylo -> Level -> String
 241 -- synchronicDistance phylo lvl =
 242 --     foldl' (\acc branch ->
 243 --              acc <> (foldl' (\acc' period ->
 244 --                               acc' <> let prox  = phyloProximity $ getConfig phylo
 245 --                                           sync  = phyloSynchrony $ getConfig phylo
 246 --                                           docs  = _phylo_timeDocs phylo
 247 --                                           prd   = _phylo_groupPeriod $ head' "distance" period
 248 --                                           edges = groupsToEdges prox 0.1 (_bpt_sensibility sync)
 249 --                                                   ((sum . elems) $ restrictKeys docs $ periodsToYears [_phylo_groupPeriod $ head' "distance" period]) period
 250 --                                       in foldl' (\mem (_,w) ->
 251 --                                           mem <> show (prd)
 252 --                                               <> "\t"
 253 --                                               <> show (w)
 254 --                                               <> "\n"
 255 --                                         ) "" edges
 256 --                      ) ""  $ elems $ groupByField _phylo_groupPeriod branch)
 257 --     ) "period\tdistance\n" $ elems $ groupByField _phylo_groupBranchId $ getGroupsFromLevel lvl phylo