{-|
Module : Gargantext.Core.Text.Ngrams
Description : Ngrams definition and tools
Copyright : (c) CNRS, 2017 - present
License : AGPL + CECILL v3
Maintainer : team@gargantext.org
Stability : experimental
Portability : POSIX
An @n-gram@ is a contiguous sequence of n items from a given sample of
text. In Gargantext application the items are words, n is a non negative
integer.
Using Latin numerical prefixes, an n-gram of size 1 is referred to as a
"unigram"; size 2 is a "bigram" (or, less commonly, a "digram"); size
3 is a "trigram". English cardinal numbers are sometimes used, e.g.,
"four-gram", "five-gram", and so on.
Source: https://en.wikipedia.org/wiki/Ngrams
TODO
group Ngrams -> Tree
compute occ by node of Tree
group occs according groups
compute cooccurrences
compute graph
-}
{-# OPTIONS_GHC -fno-warn-deprecations #-}
{-# LANGUAGE ConstrainedClassMethods #-}
{-# LANGUAGE StandaloneDeriving #-}
{-# LANGUAGE TemplateHaskell #-}
module Gargantext.Core.Text.Terms
where
import Control.Lens ( view, over )
import Data.HashMap.Strict (HashMap)
import Data.HashMap.Strict qualified as HashMap
import Data.List qualified as List
import Data.Set qualified as Set
import Data.Text qualified as Text
import GHC.Base (String)
import Gargantext.Core ( Lang, NLPServerConfig, PosTagAlgo )
import Gargantext.Core.Text (sentences, HasText(..))
import Gargantext.Core.Text.Ngrams (Ngrams(..), NgramsType(..), ngramsTerms)
import Gargantext.Core.Text.Terms.Eleve (mainEleveWith, Tries, Token, buildTries, toToken)
import Gargantext.Core.Text.Terms.Mono (monoTerms)
import Gargantext.Core.Text.Terms.Mono.Stem (stem, StemmingAlgorithm(..))
import Gargantext.Core.Text.Terms.Mono.Token.En (tokenize)
import Gargantext.Core.Text.Terms.Multi (multiterms)
import Gargantext.Core.Types ( TermsCount, POS, Terms(..), TermsWithCount )
import Gargantext.Core.Utils (groupWithCounts)
import Gargantext.Database.Prelude (DBCmd)
import Gargantext.Database.Query.Table.Ngrams (insertNgrams)
import Gargantext.Database.Query.Table.NgramsPostag (NgramsPostag(..), insertNgramsPostag, np_form, np_lem)
import Gargantext.Database.Schema.Ngrams (text2ngrams, NgramsId)
import Gargantext.Prelude
data TermType lang
= Mono { _tt_lang :: !lang }
| Multi { _tt_lang :: !lang }
| MonoMulti { _tt_lang :: !lang }
| Unsupervised { _tt_lang :: !lang
, _tt_windowSize :: !Int
, _tt_ngramsSize :: !Int
, _tt_model :: !(Maybe (Tries Token ()))
}
deriving (Generic)
deriving instance (Show lang) => Show (TermType lang)
makeLenses ''TermType
--group :: [Text] -> [Text]
--group = undefined
-- remove Stop Words
-- map (filter (\t -> not . elem t)) $
------------------------------------------------------------------------
-- | Sugar to extract terms from text (hidding 'mapM' from end user).
--extractTerms :: Traversable t => TermType Lang -> t Text -> IO (t [Terms])
extractTerms :: NLPServerConfig -> TermType Lang -> [Text] -> IO [[TermsWithCount]]
extractTerms ncs (Unsupervised {..}) xs = mapM (terms ncs (Unsupervised { _tt_model = Just m', .. })) xs
where
m' = case _tt_model of
Just m''-> m''
Nothing -> newTries _tt_windowSize (Text.unwords xs)
extractTerms ncs termTypeLang xs = mapM (terms ncs termTypeLang) xs
------------------------------------------------------------------------
withLang :: (Foldable t, Functor t, HasText h)
=> TermType Lang
-> t h
-> TermType Lang
withLang (Unsupervised {..}) ns = Unsupervised { _tt_model = m', .. }
where
m' = case _tt_model of
Nothing -> -- trace ("buildTries here" :: String)
Just $ buildTries _tt_ngramsSize
$ fmap toToken
$ uniText
$ Text.intercalate " . "
$ concatMap hasText ns
just_m -> just_m
withLang l _ = l
------------------------------------------------------------------------
data ExtractedNgrams = SimpleNgrams { unSimpleNgrams :: Ngrams }
| EnrichedNgrams { unEnrichedNgrams :: NgramsPostag }
deriving (Eq, Ord, Generic, Show)
instance Hashable ExtractedNgrams
-- | A typeclass that represents extracting ngrams from an entity.
class ExtractNgramsT h
where
extractNgramsT :: HasText h
=> NLPServerConfig
-> TermType Lang
-> h
-> DBCmd err (HashMap ExtractedNgrams (Map NgramsType Int, TermsCount))
------------------------------------------------------------------------
enrichedTerms :: Lang -> PosTagAlgo -> POS -> Terms -> NgramsPostag
enrichedTerms l pa po (Terms { .. }) =
NgramsPostag { _np_lang = l
, _np_algo = pa
, _np_postag = po
, _np_form = form
, _np_lem = lem }
where
form = text2ngrams $ Text.unwords _terms_label
lem = text2ngrams $ Text.unwords $ Set.toList _terms_stem
------------------------------------------------------------------------
cleanNgrams :: Int -> Ngrams -> Ngrams
cleanNgrams s ng
| Text.length (ng ^. ngramsTerms) < s = ng
| otherwise = text2ngrams (Text.take s (ng ^. ngramsTerms))
cleanExtractedNgrams :: Int -> ExtractedNgrams -> ExtractedNgrams
cleanExtractedNgrams s (SimpleNgrams ng) = SimpleNgrams $ cleanNgrams s ng
cleanExtractedNgrams s (EnrichedNgrams ng) = EnrichedNgrams $ over np_form (cleanNgrams s)
$ over np_lem (cleanNgrams s) ng
extracted2ngrams :: ExtractedNgrams -> Ngrams
extracted2ngrams (SimpleNgrams ng) = ng
extracted2ngrams (EnrichedNgrams ng) = view np_form ng
---------------------------
insertExtractedNgrams :: [ ExtractedNgrams ] -> DBCmd err (HashMap Text NgramsId)
insertExtractedNgrams ngs = do
let (s, e) = List.partition isSimpleNgrams ngs
m1 <- insertNgrams (map unSimpleNgrams s)
--printDebug "others" m1
m2 <- insertNgramsPostag (map unEnrichedNgrams e)
--printDebug "terms" m2
pure $ HashMap.union m1 m2
isSimpleNgrams :: ExtractedNgrams -> Bool
isSimpleNgrams (SimpleNgrams _) = True
isSimpleNgrams _ = False
------------------------------------------------------------------------
-- | Terms from 'Text'
-- 'Mono' : mono terms
-- 'Multi' : multi terms
-- 'MonoMulti' : mono and multi
-- TODO : multi terms should exclude mono (intersection is not empty yet)
terms :: NLPServerConfig -> TermType Lang -> Text -> IO [TermsWithCount]
terms _ (Mono lang) txt = pure $ monoTerms lang txt
terms ncs (Multi lang) txt = multiterms ncs lang txt
terms ncs (MonoMulti lang) txt = terms ncs (Multi lang) txt
terms _ (Unsupervised { .. }) txt = pure $ termsUnsupervised (Unsupervised { _tt_model = Just m', .. }) txt
where
m' = maybe (newTries _tt_ngramsSize txt) identity _tt_model
-- terms (WithList list) txt = pure . concat $ extractTermsWithList list txt
------------------------------------------------------------------------
type WindowSize = Int
type MinNgramSize = Int
-- | Unsupervised ngrams extraction
-- language agnostic extraction
-- TODO: newtype BlockText
termsUnsupervised :: TermType Lang -> Text -> [TermsWithCount]
termsUnsupervised (Unsupervised { _tt_model = Nothing }) = panicTrace "[termsUnsupervised] no model"
termsUnsupervised (Unsupervised { _tt_model = Just _tt_model, .. }) =
map (first (text2term _tt_lang))
. groupWithCounts
-- . List.nub
. List.filter (\l' -> List.length l' >= _tt_windowSize)
. List.concat
. mainEleveWith _tt_model _tt_ngramsSize
. uniText
termsUnsupervised _ = undefined
newTries :: Int -> Text -> Tries Token ()
newTries n t = buildTries n (toToken <$> uniText t)
-- | TODO removing long terms > 24
uniText :: Text -> [[Text]]
uniText = map (List.filter (not . isPunctuation) . tokenize)
. sentences -- TODO get sentences according to lang
. Text.toLower
text2term :: Lang -> [Text] -> Terms
text2term _ [] = Terms [] Set.empty
text2term lang txt = Terms txt (Set.fromList $ map (stem lang PorterAlgorithm) txt)
isPunctuation :: Text -> Bool
isPunctuation x = List.elem x $ Text.pack . pure
<$> ("!?(),;.:" :: String)