2 Module : Gargantext.Core.Text.Ngrams.Lists
3 Description : Tools to build lists
4 Copyright : (c) CNRS, 2017-Present
5 License : AGPL + CECILL v3
6 Maintainer : team@gargantext.org
7 Stability : experimental
12 {-# LANGUAGE ScopedTypeVariables #-}
13 {-# LANGUAGE TemplateHaskell #-}
15 module Gargantext.Core.Text.List
19 import Control.Lens ((^.), set, view, over)
21 import Data.Maybe (catMaybes)
22 import Data.Ord (Down(..))
24 import Data.Text (Text)
25 import Gargantext.API.Ngrams.Types (NgramsElement)
26 import Gargantext.API.Ngrams.Types (RepoCmdM)
27 import Gargantext.Core.Text.List.Group
28 import Gargantext.Core.Text.List.Group.Prelude
29 import Gargantext.Core.Text.List.Group.WithStem
30 import Gargantext.Core.Text.List.Social
31 import Gargantext.Core.Text.List.Social.Prelude
32 import Gargantext.Core.Text.Metrics (scored', Scored(..), normalizeGlobal, normalizeLocal)
33 import Gargantext.Core.Types (ListType(..), MasterCorpusId, UserCorpusId)
34 import Gargantext.Core.Types.Individu (User(..))
35 import Gargantext.Database.Action.Metrics.NgramsByNode (getNodesByNgramsUser, getNodesByNgramsOnlyUser)
36 import Gargantext.Database.Action.Metrics.TFICF (getTficf)
37 import Gargantext.Database.Admin.Types.Node (NodeId)
38 import Gargantext.Database.Prelude (CmdM)
39 import Gargantext.Database.Query.Table.Node (defaultList)
40 import Gargantext.Database.Query.Table.Node.Error (HasNodeError())
41 import Gargantext.Database.Query.Tree.Error (HasTreeError)
42 import Gargantext.Database.Schema.Ngrams (NgramsType(..))
43 import Gargantext.Prelude
44 import qualified Data.Char as Char
45 import qualified Data.List as List
46 import qualified Data.Map as Map
47 import qualified Data.Set as Set
48 import qualified Data.Text as Text
51 -- | TODO improve grouping functions of Authors, Sources, Institutes..
52 buildNgramsLists :: ( RepoCmdM env err m
61 -> m (Map NgramsType [NgramsElement])
62 buildNgramsLists user gp uCid mCid = do
63 ngTerms <- buildNgramsTermsList user uCid mCid gp
64 othersTerms <- mapM (buildNgramsOthersList user uCid (ngramsGroup GroupIdentity))
65 [ (Authors , MapListSize 9)
66 , (Sources , MapListSize 9)
67 , (Institutes, MapListSize 9)
70 pure $ Map.unions $ [ngTerms] <> othersTerms
73 data MapListSize = MapListSize { unMapListSize :: !Int }
75 buildNgramsOthersList ::( HasNodeError err
83 -> (NgramsType, MapListSize)
84 -> m (Map NgramsType [NgramsElement])
85 buildNgramsOthersList user uCid groupIt (nt, MapListSize mapListSize) = do
86 ngs' :: Map Text (Set NodeId) <- getNodesByNgramsUser uCid nt
88 socialLists' :: FlowCont Text FlowListScores
89 <- flowSocialList' MySelfFirst user nt (FlowCont Map.empty $ Set.fromList $ Map.keys ngs')
90 -- PrivateFirst for first developments since Public NodeMode is not implemented yet
93 printDebug "flowSocialList'"
94 $ Map.filter (not . ((==) Map.empty) . view fls_parents)
95 $ view flc_scores socialLists'
99 groupParams = GroupedTextParams groupIt (Set.size . snd) fst snd {-(size . fst)-}
100 groupedWithList = toGroupedText groupParams (view flc_scores socialLists') ngs'
103 printDebug "groupedWithList"
104 $ Map.map (\v -> (view gt_label v, view gt_children v))
105 $ Map.filter (\v -> (Set.size $ view gt_children v) > 0)
110 (stopTerms, tailTerms) = Map.partition ((== Just StopTerm) . viewListType) groupedWithList
111 (mapTerms, tailTerms') = Map.partition ((== Just MapTerm) . viewListType) tailTerms
113 listSize = mapListSize - (List.length mapTerms)
114 (mapTerms', candiTerms) = List.splitAt listSize
115 $ List.sortOn (Down . viewScore)
116 $ Map.elems tailTerms'
118 pure $ Map.fromList [( nt, (toNgramsElement stopTerms)
119 <> (toNgramsElement mapTerms )
120 <> (toNgramsElement $ setListType (Just MapTerm ) mapTerms' )
121 <> (toNgramsElement $ setListType (Just CandidateTerm) candiTerms)
126 buildNgramsTermsList :: ( HasNodeError err
135 -> m (Map NgramsType [NgramsElement])
136 buildNgramsTermsList user uCid mCid groupParams = do
138 -- Computing global speGen score
139 allTerms :: Map Text Double <- getTficf uCid mCid NgramsTerms
141 -- printDebug "head candidates" (List.take 10 $ allTerms)
142 -- printDebug "tail candidates" (List.take 10 $ List.reverse $ allTerms)
144 -- First remove stops terms
145 socialLists <- flowSocialList user NgramsTerms (Set.fromList $ map fst $ Map.toList allTerms)
146 -- printDebug "\n * socialLists * \n" socialLists
148 -- Grouping the ngrams and keeping the maximum score for label
149 let grouped = groupedTextWithStem ( GroupedTextParams (ngramsGroup groupParams) identity (const Set.empty) (const Set.empty) {-(size . _gt_label)-} ) allTerms
151 groupedWithList = map (addListType (invertForw socialLists)) grouped
153 (stopTerms, candidateTerms) = Map.partition (\t -> t ^. gt_listType == Just StopTerm) groupedWithList
154 (groupedMono, groupedMult) = Map.partition (\t -> t ^. gt_size < 2) candidateTerms
156 -- printDebug "\n * stopTerms * \n" stopTerms
157 -- splitting monterms and multiterms to take proportional candidates
159 listSizeGlobal = 2000 :: Double -- use % of list if to big, or Int if too small
160 monoSize = 0.4 :: Double
161 multSize = 1 - monoSize
163 splitAt n' ns = List.splitAt (round $ n' * listSizeGlobal) $ List.sort $ Map.elems ns
165 (groupedMonoHead, groupedMonoTail) = splitAt monoSize groupedMono
166 (groupedMultHead, groupedMultTail) = splitAt multSize groupedMult
168 -- printDebug "groupedMonoHead" (List.length groupedMonoHead)
169 -- printDebug "groupedMonoTail" (List.length groupedMonoHead)
170 -- printDebug "groupedMultHead" (List.length groupedMultHead)
171 -- printDebug "groupedMultTail" (List.length groupedMultTail)
174 -- Get Local Scores now for selected grouped ngrams
175 selectedTerms = Set.toList $ List.foldl'
176 (\set' (GroupedText _ l' _ g _ _ _ ) -> Set.union set'
180 (groupedMonoHead <> groupedMultHead)
182 -- TO remove (and remove HasNodeError instance)
183 userListId <- defaultList uCid
184 masterListId <- defaultList mCid
186 mapTextDocIds <- getNodesByNgramsOnlyUser uCid [userListId, masterListId] NgramsTerms selectedTerms
189 mapGroups = Map.fromList
190 $ map (\g -> (g ^. gt_stem, g))
191 $ groupedMonoHead <> groupedMultHead
193 -- grouping with Set NodeId
194 contextsAdded = foldl' (\mapGroups' k ->
195 let k' = ngramsGroup groupParams k in
196 case Map.lookup k' mapGroups' of
197 Nothing -> mapGroups'
198 Just g -> case Map.lookup k mapTextDocIds of
199 Nothing -> mapGroups'
200 Just ns -> Map.insert k' (over gt_nodes (Set.union ns) g) mapGroups'
203 $ Map.keys mapTextDocIds
205 -- compute cooccurrences
206 mapCooc = Map.filter (>2)
207 $ Map.fromList [ ((t1, t2), Set.size $ Set.intersection s1 s2)
208 | (t1, s1) <- mapStemNodeIds
209 , (t2, s2) <- mapStemNodeIds
210 --, t1 >= t2 -- permute byAxis diag -- since matrix symmetric
213 mapStemNodeIds = Map.toList $ Map.map (_gt_nodes) contextsAdded
214 -- printDebug "mapCooc" mapCooc
218 mapScores f = Map.fromList
219 $ map (\(Scored t g s') -> (t, f (g,s')))
224 groupsWithScores = catMaybes
226 -> case Map.lookup stem mapScores' of
228 Just s' -> Just $ g { _gt_score = s'}
229 ) $ Map.toList contextsAdded
231 mapScores' = mapScores identity
232 -- adapt2 TOCHECK with DC
233 -- printDebug "groupsWithScores" groupsWithScores
235 -- sort / partition / split
236 -- filter mono/multi again
237 (monoScored, multScored) = List.partition (\g -> _gt_size g < 2) groupsWithScores
238 -- filter with max score
239 partitionWithMaxScore = List.partition (\g -> let (s1,s2) = _gt_score g in s1 > s2 )
241 (monoScoredIncl, monoScoredExcl) = partitionWithMaxScore monoScored
242 (multScoredIncl, multScoredExcl) = partitionWithMaxScore multScored
246 listSizeLocal = 1000 :: Double -- use % of list if to big, or Int if to small
247 inclSize = 0.4 :: Double
248 exclSize = 1 - inclSize
249 splitAt' n' = List.splitAt (round $ n' * listSizeLocal)
251 (monoScoredInclHead, monoScoredInclTail) = splitAt' (monoSize * inclSize / 2) $ List.sortOn (Down . _gt_score) monoScoredIncl
252 (monoScoredExclHead, monoScoredExclTail) = splitAt' (monoSize * inclSize / 2) $ List.sortOn (Down . _gt_score) monoScoredExcl
254 (multScoredInclHead, multScoredInclTail) = splitAt' (multSize * exclSize / 2) $ List.sortOn (Down . _gt_score) multScoredIncl
255 (multScoredExclHead, multScoredExclTail) = splitAt' (multSize * exclSize / 2) $ List.sortOn (Down . _gt_score) multScoredExcl
258 -- Final Step building the Typed list
259 termListHead = maps <> cands
261 maps = set gt_listType (Just MapTerm)
262 <$> monoScoredInclHead
263 <> monoScoredExclHead
264 <> multScoredInclHead
265 <> multScoredExclHead
267 cands = set gt_listType (Just CandidateTerm)
268 <$> monoScoredInclTail
269 <> monoScoredExclTail
270 <> multScoredInclTail
271 <> multScoredExclTail
273 termListTail = map (set gt_listType (Just CandidateTerm)) ( groupedMonoTail <> groupedMultTail)
275 -- printDebug "monoScoredInclHead" monoScoredInclHead
276 -- printDebug "monoScoredExclHead" monoScoredExclTail
277 -- printDebug "multScoredInclHead" multScoredInclHead
278 -- printDebug "multScoredExclTail" multScoredExclTail
280 let result = Map.unionsWith (<>)
281 [ Map.fromList [( NgramsTerms, (List.concat $ map toNgramsElement $ termListHead)
282 <> (List.concat $ map toNgramsElement $ termListTail)
283 <> (List.concat $ map toNgramsElement $ stopTerms)
286 -- printDebug "\n result \n" r
291 toGargList :: Maybe ListType -> b -> (Maybe ListType, b)
292 toGargList l n = (l,n)
295 isStopTerm :: StopSize -> Text -> Bool
296 isStopTerm (StopSize n) x = Text.length x < n || any isStopChar (Text.unpack x)
298 isStopChar c = not (c `elem` ("- /()%" :: [Char]) || Char.isAlpha c)
300 ------------------------------------------------------------------------------