msyds/gf-core
1
0
Fork 0
forked from GitHub/gf-core
Code Pull Requests Activity

DictEngFin improved by giving preference to RGL Lexicon whenever available

Browse Source
This commit is contained in:
aarne
2012-05-25 07:52:29 +00:00
parent 5a27861038
commit 2d52a34091
2 changed files with 396 additions and 733 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1114
lib/src/finnish/DictEngFin.gf
View File

File diff suppressed because it is too large Load Diff

15
lib/src/finnish/wordnet/FreqFin.hs
View File

@@ -11,6 +11,7 @@ import System
-- AR 23/5/2012
main = do
lexicon <- readFile "../../abstract/Lexicon.gf" >>= return . getLexicon
freqs <- readFile "taajuus.txt" >>= return . getFreqMap
morpho <- readFile "../DictFin.gf" >>= return . getMorphoMap
transV <- readFile "Ven_fi.txt" >>= return . getTransDict "V" freqs morpho
@@ -18,7 +19,7 @@ main = do
transA <- readFile "Aen_fi.txt" >>= return . getTransDict "A" freqs morpho
transN <- readFile "Nen_fi.txt" >>= return . getTransDict "N" freqs morpho
transAdv <- readFile "Adven_fi.txt" >>= return . getTransDict "Adv" freqs morpho
let cnc = sort $ lmap mkLin $ transV ++ transV2 ++ transA ++ transN ++ transAdv
let cnc = sort $ lmap mkLin $ Data.List.filter (notLex lexicon) $ transV ++ transV2 ++ transA ++ transN ++ transAdv
system $ "cp dictBegin dictEngFin"
mapM_ (appendFile "dictEngFin") cnc
system $ "cat dictEngFin dictEnd >DictEngFin.gf"
@@ -41,6 +42,18 @@ getFreq ws = case ws of
n:a:r:w:c:_ -> (w,(read n,c))
-- trusted lexicon overrides all other decisions
getLexicon :: String -> MorphoMap
getLexicon = fromList . concat . lmap (getLex . words) . lines where
getLex ws = case ws of
fun:":":cat:_ -> [(takeWhile (/='_') fun, (cat,fun))]
_ -> []
notLex :: MorphoMap -> (Word,(Cat,[(Word,(Rank,Lin))])) -> Bool
notLex morpho (word,(cat,_)) = case mlookup word morpho of
Just (c,l) -> False --- | c == cat -> False --- love_V2/love_N
_ -> True
type MorphoMap = Map Word (Cat,Lin)
getMorphoMap = fromList . concat . lmap (getMorpho . words) . lines