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

stemming in Fin also for verbs now

Browse Source
This commit is contained in:
aarne
2013-03-15 11:13:49 +00:00
parent 3b66a870d1
commit 87d465594b
4 changed files with 249 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
lib/src/finnish/stemmed/ParadigmsFin.gf
View File

@@ -302,7 +302,7 @@ oper
---- mk1A : Str -> A = \jalo -> aForms2A (nforms2aforms (nForms1 jalo)) ;
---- mkNA : N -> A = snoun2sadj ;
mk1N : (talo : Str) -> N = \s -> nforms2snoun (nForms1 s) ;
mk1N : (talo : Str) -> N = \s -> lin N (nforms2snoun (nForms1 s)) ;
mk2N : (talo,talon : Str) -> N = \s,t -> nforms2snoun (nForms2 s t) ;
mk3N : (talo,talon,taloja : Str) -> N = \s,t,u -> nforms2snoun (nForms3 s t u) ;
mk4N : (talo,talon,taloa,taloja : Str) -> N = \s,t,u,v ->
@@ -537,21 +537,21 @@ oper
mkV : (
huutaa,huudan,huutaa,huutavat,huutakaa,huudetaan,
huusin,huusi,huusisi,huutanut,huudettu,huutanee : Str) -> V = mk12V ;
mkV : (sana : VK) -> V = \w -> vforms2V w.s ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
mkV : V -> Str -> V = \w,p -> vforms2V w.s ** {sc = NPCase Nom ; lock_V = <> ; p = p} ;
mkV : (sana : VK) -> V = \w -> vforms2sverb w.s ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
mkV : V -> Str -> V = \w,p -> vforms2sverb w.s ** {sc = NPCase Nom ; lock_V = <> ; p = p} ;
} ;
mk1V : Str -> V = \s ->
let vfs = vforms2V (vForms1 s) in
let vfs = vforms2sverb (vForms1 s) in
vfs ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
mk2V : (_,_ : Str) -> V = \x,y ->
let vfs = vforms2V (vForms2 x y) in vfs ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
let vfs = vforms2sverb (vForms2 x y) in vfs ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
mk3V : (huutaa,huudan,huusi : Str) -> V = \x,_,y -> mk2V x y ; ----
mk12V : (
huutaa,huudan,huutaa,huutavat,huutakaa,huudetaan,
huusin,huusi,huusisi,huutanut,huudettu,huutanee : Str) -> V =
\a,b,c,d,e,f,g,h,i,j,k,l ->
vforms2V (vForms12 a b c d e f g h i j k l) ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
vforms2sverb (vForms12 a b c d e f g h i j k l) ** {sc = NPCase Nom ; lock_V = <> ; p = []} ;
vForms1 : Str -> VForms = \ottaa ->
let
@@ -614,9 +614,11 @@ oper
caseV c v = {s = v.s ; sc = NPCase c ; qp = v.qp ; lock_V = <> ; p = v.p} ;
caseV c v = {s = v.s ; sc = NPCase c ; h = v.h ; lock_V = <> ; p = v.p} ;
vOlla = verbOlla ** {sc = NPCase Nom ; qp = True ; lock_V = <> ; p = []} ; ---- lieneekö
vOlla = {
s = table VForm ["olla";"ole";"on";"o";"olk";"olla";"oli";"oli";"olisi";"oll";"oltu";"ollu";"liene"] ;
sc = NPCase Nom ; h = Back ; lock_V = <> ; p = []} ; ---- lieneekö
mk2V2 : V -> Prep -> V2 = \v,c -> v ** {c2 = c ; lock_V2 = <>} ;
caseV2 : V -> Case -> V2 = \v,c -> mk2V2 v (casePrep c) ;
@@ -632,7 +634,7 @@ oper
mkV2 : V -> V2 = dirV2 ;
mkV2 : V -> Case -> V2 = caseV2 ;
mkV2 : V -> Prep -> V2 = mk2V2 ;
mkV2 : VK -> V2 = \w -> dirV2 (vforms2V w.s ** {sc = NPCase Nom ; lock_V = <> ; p = []}) ;
mkV2 : VK -> V2 = \w -> dirV2 (vforms2sverb w.s ** {sc = NPCase Nom ; lock_V = <> ; p = []}) ;
} ;
mk2V2 : V -> Prep -> V2 ;

67
lib/src/finnish/stemmed/SentenceFin.gf Normal file
View File

@@ -0,0 +1,67 @@
concrete SentenceFin of Sentence = CatFin ** open Prelude, ResFin, StemFin in {
flags optimize=all_subs ;
lin
PredVP np vp = mkClausePol (orB np.isNeg vp.isNeg) (subjForm np vp.sc) np.a vp ;
PredSCVP sc vp = mkClause (\_ -> sc.s) (agrP3 Sg) vp ;
ImpVP vp = {
s = \\pol,agr =>
let
verb = vp.s ! VIImper ! Simul ! pol ! agr ;
compl = vp.s2 ! False ! pol ! agr ++ vp.ext --- False = like inf (osta auto)
in
verb.fin ++ verb.inf ++ compl ;
} ;
-- The object case is formed at the use site of $c2$, in $Relative$ and $Question$.
SlashVP np vp = {
s = \\t,a,p => (mkClause (subjForm np vp.sc) np.a vp).s ! t ! a ! p ! SDecl ;
c2 = vp.c2
} ;
AdvSlash slash adv = {
s = \\t,a,b => slash.s ! t ! a ! b ++ adv.s ;
c2 = slash.c2
} ;
SlashPrep cl prep = {
s = \\t,a,p => cl.s ! t ! a ! p ! SDecl ;
c2 = prep
} ;
SlashVS np vs slash = {
s = \\t,a,p =>
(mkClause (subjForm np vs.sc) np.a
(insertExtrapos ("että" ++ slash.s)
(predSV vs))
).s ! t ! a ! p ! SDecl ;
c2 = slash.c2
} ;
EmbedS s = {s = etta_Conj ++ s.s} ;
EmbedQS qs = {s = qs.s} ;
EmbedVP vp = {s = infVP (NPCase Nom) Pos (agrP3 Sg) vp Inf1} ; --- case,pol,agr,infform
UseCl t p cl = {s = t.s ++ p.s ++ cl.s ! t.t ! t.a ! p.p ! SDecl} ;
UseQCl t p cl = {s = t.s ++ p.s ++ cl.s ! t.t ! t.a ! p.p} ;
UseRCl t p cl = {
s = \\r => t.s ++ p.s ++ cl.s ! t.t ! t.a ! p.p ! r ;
c = cl.c
} ;
UseSlash t p cl = {
s = t.s ++ p.s ++ cl.s ! t.t ! t.a ! p.p ;
c2 = cl.c2
} ;
AdvS a s = {s = a.s ++ s.s} ;
ExtAdvS a s = {s = a.s ++ "," ++ s.s} ;
RelS s r = {s = s.s ++ "," ++ r.s ! agrP3 Sg} ; ---- mikä
}

152
lib/src/finnish/stemmed/StemFin.gf
View File

@@ -126,4 +126,156 @@ oper
} ;
-- verbs
oper
SVForm : Type = Predef.Ints 12 ;
SVerb : Type = {s : SVForm => Str ; h : Harmony} ;
-- used in Cat
SVerb1 = {s : SVForm => Str ; sc : NPForm ; h : Harmony ; p : Str} ;
sverb2verbBind : SVerb -> Verb = sverb2verb True ;
sverb2verbSep : SVerb -> Verb = sverb2verb False ;
vforms2sverb : VForms -> SVerb = \vf -> {
s = table {
0 => (vf ! 0) ; -- tulla
1 => Predef.tk 1 (vf ! 1) ; -- tule(n)
2 => (vf ! 2) ; -- tulee
3 => Predef.tk 3 (vf ! 3) ; -- tule(vat)
4 => Predef.tk 2 (vf ! 4) ; -- tulk(aa)
5 => Predef.tk 2 (vf ! 5) ; -- tulla(an)
6 => Predef.tk 1 (vf ! 6) ; -- tuli(n)
7 => (vf ! 7) ; -- tuli
8 => (vf ! 8) ; -- tulisi
9 => Predef.tk 2 (vf ! 9) ; -- tull(ut)
10 => (vf ! 10) ; -- tultu
11 => weakGrade (vf ! 10) ; -- tullu(n)
12 => Predef.tk 1 (vf ! 11) -- tulle(e)
} ;
h = aHarmony (last (vf ! 0)) ;
} ;
sverb2verb : Bool -> SVerb -> Verb = \b,sverb ->
let
plus = plusIf b ;
vh = sverb.s ;
tulla = vh ! 0 ;
tule_ = vh ! 1 ; -- tule(n)
tulee = vh ! 2 ;
tule__ = vh ! 3 ; -- tule(vat)
tulk_ = vh ! 4 ; -- tulk(aa)
tulla_ = vh ! 5 ; -- tulla(an)
tuli_ = vh ! 6 ; -- tuli(n)
tuli = vh ! 7 ;
tulisi = vh ! 8 ;
tull_ = vh ! 9 ; -- tull(ut)
tultu = vh ! 10 ;
tullu__ = vh ! 11 ; -- tullu(n)
tulle_ = vh ! 12 ; -- tulle(e)
a = harmonyA sverb.h ;
o = harmonyV "o" "ö" sverb.h ;
u = harmonyV "u" "y" sverb.h ;
tulko = plus tulk_ o ;
tulkoo = plus tulko o ;
tullee = plus tull_ "ee" ;
tulle = plus tull_ "e" ;
tullu_ = plus tull_ u ;
tullut = plus tullu_ "t" ;
tullun = plus tullu_ "n" ;
tullutN : Noun = snoun2noun b {
s = table SNForm [
tullut ;
tullee ;
plus tullut ("t" + a) ;
plus tullee ("n" + a) ;
plus tullee "see" ;
plus tulle "ide" ;
plus tulle ("it" + a) ;
plus tulle "i" ;
plus tulle "i" ;
plus tulle "isii"
] ;
h = sverb.h
} ;
tultuN : Noun = snoun2noun b {
s = table SNForm [
tultu ;
tullu__ ;
plus tultu a ;
plus tultu ("n" + a) ;
plus tultu u ;
plus tultu "je" ;
plus tultu ("j" + a) ;
plus tultu "i" ;
plus tullu__ "i" ;
plus tultu "ihi"
] ;
h = sverb.h
} ;
tulema = plus tule__ ("m" + a) ;
vat = "v" + a + "t"
in
{s = table {
Inf Inf1 => tulla ;
Presn Sg P1 => plus tule_ "n" ;
Presn Sg P2 => plus tule_ "t" ;
Presn Sg P3 => tulee ;
Presn Pl P1 => plus tule_ "mme" ;
Presn Pl P2 => plus tule_ "tte" ;
Presn Pl P3 => plus tule__ vat ;
Impf Sg P1 => plus tuli_ "n" ; --# notpresent
Impf Sg P2 => plus tuli_ "t" ; --# notpresent
Impf Sg P3 => tuli ; --# notpresent
Impf Pl P1 => plus tuli_ "mme" ; --# notpresent
Impf Pl P2 => plus tuli_ "tte" ; --# notpresent
Impf Pl P3 => plus tuli vat ; --# notpresent
Condit Sg P1 => plus tulisi "n" ; --# notpresent
Condit Sg P2 => plus tulisi "t" ; --# notpresent
Condit Sg P3 => tulisi ; --# notpresent
Condit Pl P1 => plus tulisi "mme" ; --# notpresent
Condit Pl P2 => plus tulisi "tte" ; --# notpresent
Condit Pl P3 => plus tulisi vat ; --# notpresent
Imper Sg => tule_ ;
Imper Pl => plus tulk_ (a + a) ;
ImperP3 Sg => plus tulkoo "n" ;
ImperP3 Pl => plus tulkoo "t" ;
ImperP1Pl => plus tulk_ (a + a + "mme") ;
ImpNegPl => tulko ;
Pass True => plus tulla_ (a + "n") ;
Pass False => tulla_ ;
PastPartAct (AN n) => tullutN.s ! n ;
PastPartAct AAdv => plus tullee "sti" ;
PastPartPass (AN n) => tultuN.s ! n ;
PastPartPass AAdv => plus tullu__ "sti" ;
Inf Inf3Iness => plus tulema ("ss" + a) ;
Inf Inf3Elat => plus tulema ("st" + a) ;
Inf Inf3Illat => plus tulema (a + "n") ;
Inf Inf3Adess => plus tulema ("ll" + a) ;
Inf Inf3Abess => plus tulema ("tt" + a)
} ;
sc = NPCase Nom ;
lock_V = <>
} ;
predSV : SVerb1 -> VP = \sv ->
predV (sverb2verbSep sv ** {p = sv.p ; sc = sv.sc ; qp = case sv.h of {Back => True ; Front => False}}) ;
-- (Verb ** {sc : NPForm ; qp : Bool ; p : Str}) -> VP = \verb -> {
-- auxiliary
plusIf : Bool -> Str -> Str -> Str = \b,x,y -> case b of {
True => x + y ;
False => glue x y
} ;
}

38
lib/src/finnish/stemmed/VerbFin.gf
View File

@@ -1,54 +1,54 @@
--1 Verb Phrases in Finnish
concrete VerbFin of Verb = CatFin ** open Prelude, ResFin in {
concrete VerbFin of Verb = CatFin ** open Prelude, ResFin, StemFin in {
flags optimize=all_subs ;
lin
UseV = predV ;
UseV = predSV ;
SlashV2a v = predV v ** {c2 = v.c2} ;
SlashV2a v = predSV v ** {c2 = v.c2} ;
Slash2V3 v np =
insertObj
(\\fin,b,_ => appCompl fin b v.c2 np) (predV v) ** {c2 = v.c3} ;
(\\fin,b,_ => appCompl fin b v.c2 np) (predSV v) ** {c2 = v.c3} ;
Slash3V3 v np =
insertObj
(\\fin,b,_ => appCompl fin b v.c3 np) (predV v) ** {c2 = v.c2} ;
(\\fin,b,_ => appCompl fin b v.c3 np) (predSV v) ** {c2 = v.c2} ;
ComplVV v vp =
insertObj
(\\_,b,a => infVP v.sc b a vp v.vi)
(predV {s = v.s ;
(predSV {s = v.s ;
sc = case vp.sc of {
NPCase Nom => v.sc ; -- minun täytyy pestä auto
c => c -- minulla täytyy olla auto
} ;
qp = v.qp ; p = v.p
h = v.h ; p = v.p
}
) ;
ComplVS v s = insertExtrapos (etta_Conj ++ s.s) (predV v) ;
ComplVQ v q = insertExtrapos ( q.s) (predV v) ;
ComplVS v s = insertExtrapos (etta_Conj ++ s.s) (predSV v) ;
ComplVQ v q = insertExtrapos ( q.s) (predSV v) ;
ComplVA v ap =
insertObj
(\\_,b,agr =>
let n = (complNumAgr agr) in
ap.s ! False ! (NCase n (npform2case n v.c2.c))) --- v.cs.s ignored
(predV v) ;
(predSV v) ;
SlashV2S v s =
insertExtrapos (etta_Conj ++ s.s) (predV v) ** {c2 = v.c2} ;
insertExtrapos (etta_Conj ++ s.s) (predSV v) ** {c2 = v.c2} ;
SlashV2Q v q =
insertExtrapos (q.s) (predV v) ** {c2 = v.c2} ;
insertExtrapos (q.s) (predSV v) ** {c2 = v.c2} ;
SlashV2V v vp =
insertObj (\\_,b,a => infVP v.sc b a vp v.vi) (predV v) ** {c2 = v.c2} ;
insertObj (\\_,b,a => infVP v.sc b a vp v.vi) (predSV v) ** {c2 = v.c2} ;
---- different infinitives
SlashV2A v ap =
insertObj
(\\fin,b,_ =>
ap.s ! False ! (NCase Sg (npform2case Sg v.c3.c))) ----agr to obj
(predV v) ** {c2 = v.c2} ;
(predSV v) ** {c2 = v.c2} ;
ComplSlash vp np = insertObjPre np.isNeg (\\fin,b,_ => appCompl fin b vp.c2 np) vp ;
@@ -60,12 +60,12 @@ concrete VerbFin of Verb = CatFin ** open Prelude, ResFin in {
SlashVV v vp =
insertObj
(\\_,b,a => infVP v.sc b a vp v.vi)
(predV {s = v.s ;
(predSV {s = v.s ;
sc = case vp.sc of {
NPCase Nom => v.sc ; -- minun täytyy pestä auto
c => c -- minulla täytyy olla auto
} ;
qp = v.qp ; p = v.p
h = v.h ; p = v.p
}
) ** {c2 = vp.c2} ; ---- correct ??
{---- 153543936 (210912,312)
@@ -73,7 +73,7 @@ concrete VerbFin of Verb = CatFin ** open Prelude, ResFin in {
insertObjPre np.isNeg
(\\fin,b,a => appCompl True b v.c2 np ++ ---- fin -> stack overflow
infVP v.sc b a vp v.vi)
(predV v) ** {c2 = vp.c2} ;
(predSV v) ** {c2 = vp.c2} ;
----}
AdvVP vp adv = insertAdv (\\_ => adv.s) vp ;
@@ -86,12 +86,12 @@ concrete VerbFin of Verb = CatFin ** open Prelude, ResFin in {
ReflVP v = insertObjPre False (\\fin,b,agr => appCompl fin b v.c2 (reflPron agr)) v ;
PassV2 v = let vp = predV v in {
PassV2 v = let vp = predSV v in {
s = \\_ => vp.s ! VIPass ;
s2 = \\_,_,_ => [] ;
adv = \\_ => [] ;
ext = [] ;
qp = v.qp ;
qp = vp.qp ;
isNeg = False ;
sc = v.c2.c -- minut valitaan ; minua rakastetaan ; minulle kuiskataan
} ; ---- talon valitaan: should be marked like inf.