fast-unidic / feature.def
andito's picture
andito HF staff
Upload folder using huggingface_hub
65e2480 verified
# F[0]: pos1
# F[1]: pos2
# F[2]: pos3
# F[3]: pos4
# F[4]: cType
# F[5]: cForm
# F[6]: lForm
# F[7]: lemma
# F[8]: orth
# F[9]: orthBase
# F[10]: pron
# F[11]: pronBase
# F[12]: goshu
# F[13]: aType
# F[14]: aConType
# F[15]: aModType
#
# L[0]: pos1
# L[1]: pos2
# L[2]: pos3
# L[3]: pos4
# L[4]: cType
# L[5]: cForm
# L[6]: orth
# L[7]: orthBase
# L[8]: goshu
# L[9]: fType
# L[10]: fForm
# L[11]: iConType
# L[12]: aType
# L[13]: aConType
# L[14]: aModType
#
# R[0]: pos1
# R[1]: pos2
# R[2]: pos3
# R[3]: pos4
# R[4]: cType
# R[5]: cForm
# R[6]: orth
# R[7]: orthBase
# R[8]: goshu
# R[9]: iType
# R[10]: iForm
# R[11]: fConType
# R[12]: aType
# R[13]: aConType
# R[14]: aModType
UNIGRAM LGOP01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3],%F?[8],%F?[10]
UNIGRAM LGOP02:%F[6],%F[7],%F[0],%F[1],%F?[2],%F?[8],%F?[10]
UNIGRAM LGOP03:%F[6],%F[7],%F[0],%F?[1],%F?[8],%F?[10]
UNIGRAM LGOP03:%F[6],%F[7],%F[0],%F?[8],%F?[10]
UNIGRAM LGCOP01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[5],%F?[8],%F?[10]
UNIGRAM LGCOP02:%F[6],%F[7],%F[0],%F?[4],%F?[5],%F?[8],%F?[10]
UNIGRAM LGO01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3],%F?[8]
UNIGRAM LGO02:%F[6],%F[7],%F[0],%F[1],%F?[2],%F?[8]
UNIGRAM LGO03:%F[6],%F[7],%F[0],%F?[1],%F?[8]
UNIGRAM LGO03:%F[6],%F[7],%F[0],%F?[8]
UNIGRAM LGCO01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[5],%F?[8]
UNIGRAM LGCO02:%F[6],%F[7],%F[0],%F?[4],%F?[5],%F?[8]
UNIGRAM LG01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3]
UNIGRAM LG02:%F[6],%F[7],%F[0],%F[1],%F?[2]
UNIGRAM LG03:%F[6],%F[7],%F[0],%F?[1]
UNIGRAM LG03:%F[6],%F[7],%F[0]
UNIGRAM LGC01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[5]
UNIGRAM LGC02:%F[6],%F[7],%F[0],%F?[4],%F?[5]
UNIGRAM LGCt01:%F[6],%F[7],%F[0],%F?[1],%F?[4]
UNIGRAM LGCt02:%F[6],%F[7],%F[0],%F?[4]
UNIGRAM L01:%F[6],%F[7]
UNIGRAM LGP01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3],%F?[10]
UNIGRAM LGP02:%F[6],%F[7],%F[0],%F[1],%F?[2],%F?[10]
UNIGRAM LGP03:%F[6],%F[7],%F[0],%F?[1],%F?[10]
UNIGRAM LGP03:%F[6],%F[7],%F[0],%F?[10]
UNIGRAM LGCP01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[5],%F?[10]
UNIGRAM LGCP02:%F[6],%F[7],%F[0],%F?[4],%F?[5],%F?[10]
UNIGRAM GC01:%F[0],%F?[4]
UNIGRAM GC02:%F[0],%F?[1],%F?[4]
UNIGRAM GC03:%F[0],%F?[4],%F?[5]
UNIGRAM GC04:%F[0],%F?[1],%F?[4],%F?[5]
UNIGRAM G01:%F[0]
UNIGRAM G02:%F[0],%F?[1]
UNIGRAM G03:%F[0],%F[1],%F?[2]
UNIGRAM G04:%F[0],%F[1],%F[2],%F?[3]
UNIGRAM C01:%F?[4]
UNIGRAM C03:%F?[4],%F?[5]
UNIGRAM O01:%F?[8]
UNIGRAM O02:%F?[9]
UNIGRAM P01:%F?[10]
UNIGRAM P02:%F?[11]
UNIGRAM W01:%F?[12]
UNIGRAM WG01:%F?[12],%F[0]
UNIGRAM T01:%t
UNIGRAM TG01:%t,%F[0]
UNIGRAM GOP01:%F[0],%F[1],%F[2],%F?[3],%F?[8],%F?[10]
UNIGRAM GOP02:%F[0],%F[1],%F?[2],%F?[8],%F?[10]
UNIGRAM GOP03:%F[0],%F?[1],%F?[8],%F?[10]
UNIGRAM GOP03:%F[0],%F?[8],%F?[10]
UNIGRAM GCOP01:%F[0],%F?[1],%F?[4],%F?[5],%F?[8],%F?[10]
UNIGRAM GCOP02:%F[0],%F?[4],%F?[5],%F?[8],%F?[10]
UNIGRAM GO01:%F[0],%F[1],%F[2],%F?[3],%F?[8]
UNIGRAM GO02:%F[0],%F[1],%F?[2],%F?[8]
UNIGRAM GO03:%F[0],%F?[1],%F?[8]
UNIGRAM GO03:%F[0],%F?[8]
UNIGRAM GCO01:%F[0],%F?[1],%F?[4],%F?[5],%F?[8]
UNIGRAM GCO02:%F[0],%F?[4],%F?[5],%F?[8]
UNIGRAM GP01:%F[0],%F[1],%F[2],%F?[3],%F?[10]
UNIGRAM GP02:%F[0],%F[1],%F?[2],%F?[10]
UNIGRAM GP03:%F[0],%F?[1],%F?[10]
UNIGRAM GP03:%F[0],%F?[10]
UNIGRAM GCP01:%F[0],%F?[1],%F?[4],%F?[5],%F?[10]
UNIGRAM GCP02:%F[0],%F?[4],%F?[5],%F?[10]
UNIGRAM LGObPb01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3],%F?[9],%F?[11]
UNIGRAM LGObPb02:%F[6],%F[7],%F[0],%F[1],%F?[2],%F?[9],%F?[11]
UNIGRAM LGObPb03:%F[6],%F[7],%F[0],%F?[1],%F?[9],%F?[11]
UNIGRAM LGObPb03:%F[6],%F[7],%F[0],%F?[9],%F?[11]
UNIGRAM LGCObPb01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[9],%F?[11]
UNIGRAM LGCObPb02:%F[6],%F[7],%F[0],%F?[4],%F?[9],%F?[11]
UNIGRAM LGOb01:%F[6],%F[7],%F[0],%F[1],%F[2],%F?[3],%F?[9]
UNIGRAM LGOb02:%F[6],%F[7],%F[0],%F[1],%F?[2],%F?[9]
UNIGRAM LGOb03:%F[6],%F[7],%F[0],%F?[1],%F?[9]
UNIGRAM LGOb03:%F[6],%F[7],%F[0],%F?[9]
UNIGRAM LGCOb01:%F[6],%F[7],%F[0],%F?[1],%F?[4],%F?[9]
UNIGRAM LGCOb02:%F[6],%F[7],%F[0],%F?[4],%F?[9]
UNIGRAM GObPb01:%F[0],%F[1],%F[2],%F?[3],%F?[9],%F?[11]
UNIGRAM GObPb02:%F[0],%F[1],%F?[2],%F?[9],%F?[11]
UNIGRAM GObPb03:%F[0],%F?[1],%F?[9],%F?[11]
UNIGRAM GObPb03:%F[0],%F?[9],%F?[11]
UNIGRAM GCObPb01:%F[0],%F?[1],%F?[4],%F?[9],%F?[11]
UNIGRAM GCObPb02:%F[0],%F?[4],%F?[9],%F?[11]
UNIGRAM GOb01:%F[0],%F[1],%F[2],%F?[3],%F?[9]
UNIGRAM GOb02:%F[0],%F[1],%F?[2],%F?[9]
UNIGRAM GOb03:%F[0],%F?[1],%F?[9]
UNIGRAM GOb03:%F[0],%F?[9]
UNIGRAM GCOb01:%F[0],%F?[1],%F?[4],%F?[9]
UNIGRAM GCOb02:%F[0],%F?[4],%F?[9]
UNIGRAM GPb01:%F[0],%F[1],%F[2],%F?[3],%F?[11]
UNIGRAM GPb02:%F[0],%F[1],%F?[2],%F?[11]
UNIGRAM GPb03:%F[0],%F?[1],%F?[11]
UNIGRAM GPb03:%F[0],%F?[11]
UNIGRAM GCPb01:%F[0],%F?[1],%F?[4],%F?[11]
UNIGRAM GCPb02:%F[0],%F?[4],%F?[11]
UNIGRAM CA01:%F?[5],%F?[15]
############## 統語的連接素性 #################
BIGRAM G_G01:%L[0]/%R[0]
BIGRAM G_G02:%L[0]/%R[0],%R?[1]
BIGRAM G_G03:%L[0]/%R[0],%R[1],%R?[2]
BIGRAM G_G04:%L[0]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM G_G05:%L[0],%L?[1]/%R[0]
BIGRAM G_G06:%L[0],%L?[1]/%R[0],%R?[1]
BIGRAM G_G07:%L[0],%L?[1]/%R[0],%R[1],%R?[2]
BIGRAM G_G08:%L[0],%L?[1]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM G_G09:%L[0],%L[1],%L?[2]/%R[0]
BIGRAM G_G10:%L[0],%L[1],%L?[2]/%R[0],%R?[1]
BIGRAM G_G11:%L[0],%L[1],%L?[2]/%R[0],%R[1],%R?[2]
BIGRAM G_G12:%L[0],%L[1],%L?[2]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM G_G13:%L[0],%L[1],%L[2],%L?[3]/%R[0]
BIGRAM G_G14:%L[0],%L[1],%L[2],%L?[3]/%R[0],%R?[1]
BIGRAM G_G15:%L[0],%L[1],%L[2],%L?[3]/%R[0],%R[1],%R?[2]
BIGRAM G_G16:%L[0],%L[1],%L[2],%L?[3]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM G_O01:%L[0]/%R?[7]
BIGRAM G_O02:%L[0],%L?[1]/%R?[7]
BIGRAM G_O03:%L[0],%L[1],%L?[2]/%R?[7]
BIGRAM G_O04:%L[0],%L[1],%L[2],%L?[3]/%R?[7]
BIGRAM G_O05:%L[0]/%R?[6]
BIGRAM G_O06:%L[0],%L?[1]/%R?[6]
BIGRAM G_O07:%L[0],%L[1],%L?[2]/%R?[6]
BIGRAM G_O08:%L[0],%L[1],%L[2],%L?[3]/%R?[6]
#右のGはいらない汎化を防ぐため
BIGRAM G_C01:%L[0]/%R[0],%R?[4]
BIGRAM G_C02:%L[0],%L?[1]/%R[0],%R?[4]
BIGRAM G_C03:%L[0]/%R[0],%R?[4],%R[5]
BIGRAM G_C04:%L[0],%L?[1]/%R[0],%R?[4],%R[5]
BIGRAM O_G01:%L?[7]/%R[0]
BIGRAM O_G02:%L?[7]/%R[0],%R?[1]
BIGRAM O_G03:%L?[7]/%R[0],%R[1],%R?[2]
BIGRAM O_G04:%L?[7]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM O_G05:%L?[6]/%R[0]
BIGRAM O_G06:%L?[6]/%R[0],%R?[1]
BIGRAM O_G07:%L?[6]/%R[0],%R[1],%R?[2]
BIGRAM O_G08:%L?[6]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM O_O01:%L?[6]/%R?[6]
BIGRAM O_O02:%L?[6]/%R?[7]
BIGRAM O_O03:%L?[7]/%R?[7]
#右のGはいらない汎化を防ぐため
BIGRAM O_C01:%L?[6]/%R[0],%R?[4]
BIGRAM O_C02:%L?[6]/%R[0],%R?[4],%R?[5]
BIGRAM O_C03:%L?[7]/%R[0],%R?[4]
BIGRAM O_C04:%L?[7]/%R[0],%R?[4],%R?[5]
#左のGはいらない汎化を防ぐため
BIGRAM C_G01:%L[0],%L?[4]/%R[0]
BIGRAM C_G02:%L[0],%L?[4]/%R[0],%R?[1]
BIGRAM C_G03:%L[0],%L?[4]/%R[0],%R[1],%R?[2]
BIGRAM C_G04:%L[0],%L?[4]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM C_G05:%L[0],%L?[4],%L?[5]/%R[0]
BIGRAM C_G06:%L[0],%L?[4],%L?[5]/%R[0],%R?[1]
BIGRAM C_G07:%L[0],%L?[4],%L?[5]/%R[0],%R[1],%R?[2]
BIGRAM C_G08:%L[0],%L?[4],%L?[5]/%R[0],%R[1],%R[2],%R?[3]
#左のGはいらない汎化を防ぐため
BIGRAM C_O01:%L[0],%L?[4]/%R?[6]
BIGRAM C_O02:%L[0],%L?[4]/%R?[7]
BIGRAM C_O03:%L[0],%L?[4],%L?[5]/%R?[6]
BIGRAM C_O04:%L[0],%L?[4],%L?[5]/%R?[7]
#Gはいらない汎化を防ぐため
BIGRAM C_C01:%L[0],%L?[4]/%R[0],%R?[4]
BIGRAM C_C02:%L[0],%L?[4],%L?[5]/%R[0],%R?[4]
BIGRAM C_C03:%L[0],%L?[4]/%R[0],%R?[4]%R?[5]
BIGRAM C_C04:%L[0],%L?[4],%L?[5]/%R[0],%R?[4]%R?[5]
#L[0]は詳細化
BIGRAM Cf_G01:%L?[5]/%R[0]
BIGRAM Cf_G02:%L?[5]/%R[0],%R?[1]
BIGRAM Cf_G03:%L?[5]/%R[0],%R[1],%R?[2]
BIGRAM Cf_G04:%L?[5]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM Cf_G01:%L[0],%L?[5]/%R[0]
BIGRAM Cf_G02:%L[0],%L?[5]/%R[0],%R?[1]
BIGRAM Cf_G03:%L[0],%L?[5]/%R[0],%R[1],%R?[2]
BIGRAM Cf_G04:%L[0],%L?[5]/%R[0],%R[1],%R[2],%R?[3]
BIGRAM I_I %L?[11]/%R?[9],%R?[10]
BIGRAM F_F %L?[9],%L?[10]/%R?[11]
############## 語彙的連接素性 #################
BIGRAM W_W01:%L?[8]/%R?[8]
BIGRAM C_C01:%L?[4]/%R?[4]
BIGRAM W_C01:%L?[8]/%R?[4]
BIGRAM C_W01:%L?[4]/%R?[8]
BIGRAM W_O01:%L?[8]/%R?[6]
BIGRAM W_O02:%L?[8]/%R?[7]
BIGRAM O_W01:%L?[6]/%R?[8]
BIGRAM O_W02:%L?[7]/%R?[8]
BIGRAM GC_GC01:%L[0],%L?[1],%L?[4],%L?[5]/%R[0],%R?[1],%R?[4],%R?[5]
BIGRAM GC_GC02:%L[0],%L?[4],%L?[5]/%R[0],%R?[1],%R?[4],%R?[5]
BIGRAM GC_GC03:%L[0],%L?[1],%L?[4],%L?[5]/%R[0],%R?[4],%R?[5]
BIGRAM GC_GC04:%L[0],%L?[4],%L?[5]/%R[0],%R?[4],%R?[5]
############### アクセント連接素性 #################
BIGRAM A_A01:%L?[12]/%R?[12]
BIGRAM A_A02:%L?[5],%L?[14]/%R?[12]
BIGRAM A_A03:%L?[12]/%R?[5],%R?[14]
BIGRAM A_A04:%L?[5],%L?[14]/%R?[5],%R?[14]
BIGRAM G_Ac01:%L[0]/%R?[13]
BIGRAM G_Ac02:%L[0],%L?[1]/%R?[13]
BIGRAM G_Ac03:%L[0],%L?[1],%L?[2]/%R?[13]
BIGRAM Ac_G01:%L?[13]/%R[0]
BIGRAM Ac_G02:%L?[13]/%R[0],%R?[1]
BIGRAM Ac_G03:%L?[13]/%R[0],%R?[1],%R?[2]