Merge branch 'master' of https://gitlab.lif.univ-mrs.fr/alexis.nasr/maca_data1

ftb/Makefile

@@ -14,6 +14,7 @@ compile:
 	$(MAKE) -C maca_lexer compile
 	$(MAKE) -C maca_trans_tagger compile
 	$(MAKE) -C maca_trans_morpho compile
+	$(MAKE) -C maca_trans_lemmatizer compile
 	$(MAKE) -C maca_trans_parser compile
 #	$(MAKE) -C maca_crf_tagger compile
 
@@ -22,13 +23,14 @@ install:
 	$(MAKE) -C maca_lexer install
 	$(MAKE) -C maca_trans_tagger install
 	$(MAKE) -C maca_trans_morpho install
-	$(MAKE) -C maca_lemmatizer install
+	$(MAKE) -C maca_trans_lemmatizer install
 	$(MAKE) -C maca_trans_parser install
 #	$(MAKE) -C maca_crf_tagger install
 #	@tar -cvzf ./maca_datas.tgz bin
 
 evaluation:
-	$(MAKE) LANGUAGE=$(_LANG) CODE=$(_LANG) -C eval
+#	$(MAKE) LANGUAGE=$(_LANG) CODE=$(_LANG) -C eval
+	$(MAKE) -C eval
 
 clean:
 	$(MAKE) -C data/morpho-lexicon clean

ftb/data/morpho-lexicon/Makefile

-#DATA_DIR=../../../data/ud-treebanks-conll2017/UD_French
-#TRAIN=$(DATA_DIR)/fr-ud-train.conllu
+DATA_DIR=../../../data/ftb
+TRAIN=$(DATA_DIR)/ftb.train.conll07
 
 include ../../../makefiles/morpho-lexicon_ud.makefile

ftb/data/treebank/Makefile

-#DATA_DIR=../../../data/ud-treebanks-conll2017/UD_French
-#TRAIN=$(DATA_DIR)/fr-ud-train.conllu
-#TEST=$(DATA_DIR)/fr-ud-dev.conllu
-#DEV=$(DATA_DIR)/fr-ud-dev.conllu
+DATA_DIR=../../../data/ftb
+TRAIN=$(DATA_DIR)/ftb.train.conll07
+TEST=$(DATA_DIR)/ftb.test.conll07
+DEV=$(DATA_DIR)/ftb.dev.conll07
 
 include ../../../makefiles/treebank_ud.makefile
 

ftb/eval/Makefile

 
-#LANGUAGE=UD_French
+LANGUAGE=ftb
 WPMLGFS_MCD_FILE=../../mcd/wpmlgfs.mcd
 WPLGFS_MCD_FILE=../../mcd/wplgfs.mcd
 TEST=../data/treebank/test.mcf

ftb/maca_lemmatizer/Makefile

-DATA=../data/morpho-lexicon
-
-##-----------------------------------------------------------------------
-## compile
-##-----------------------------------------------------------------------
-
-##-----------------------------------------------------------------------
-## install
-##-----------------------------------------------------------------------
-
-install:
-	- cp $(DATA)/fplm ../bin
-
-##-----------------------------------------------------------------------
-## clean
-##-----------------------------------------------------------------------
-
-

ftb/maca_trans_lemmatizer/Makefile

+MCF_TRAIN=../data/treebank/train.mcf
+MCF_DEV=../data/treebank/dev.mcf
+MCF_TEST=../data/treebank/test.mcf
+CFF_TRAIN=train.cff
+MODEL_FILENAME=maca_trans_lemmatizer.model 
+RULES_FILENAME=maca_trans_lemmatizer_rules.txt
+EXCEPTIONS_FPLM_FILENAME=maca_trans_lemmatizer_exceptions.fplm
+VOCABS_FILENAME=maca_trans_lemmatizer.vocab 
+FEATURES_MODEL_FILENAME=maca_trans_lemmatizer.fm
+FPLM_FILENAME=../data/morpho-lexicon/fplm
+THRESHOLD=50
+STRICT=-s
+
+
+include ../../makefiles/maca_trans_lemmatizer.makefile

ftb/maca_trans_lemmatizer/maca_trans_lemmatizer.fm

+#b0U1
+#b0p b0f
+#b0len
+#bm1f
+#bm2f
+#b0p
+#bm1p
+#bm1m
+#bm1p bm1m
+#bm2p
+#bm3p
+#bm2p bm1p
+#bm2p bm3p
+#b0p b0s1
+#b0s2
+#b0s3
+#b0s4
+#b0s5
+b0p b0s1 b0s2
+b0p b0s2 b0s3
+b0p b0s3 b0s4
+b0p b0s4 b0s5
+b0p b0s5 b0s6
+b0p b0s1 b0s2
+b0p b0s1 b0s2 b0s3
+b0p b0s1 b0s2 b0s3 b0s4
+b0p b0s1 b0s2 b0s3 b0s4 b0s5
+b0p b0s1 b0s2 b0s3 b0s4 b0s5 b0s6

makefiles/morpho-lexicon_ud.makefile

@@ -3,7 +3,8 @@ TOOLS=../../../tools/
 compile: fplm fP
 
 fplm:
-	$(TOOLS)conll2fplm.pl -c -f $(TRAIN) > $@
+	$(TOOLS)conll2fplm.pl -f $(TRAIN) > $@
+#	$(TOOLS)conll2fplm.pl -c -f $(TRAIN) > $@
 
 fP: fplm
 	$(TOOLS)fplm2fP_ud.pl $< > $@

makefiles/treebank_ud.makefile

@@ -5,16 +5,16 @@ compile: train.mcf  test.mcf dev.mcf
 
 train.mcf: $(TRAIN)
 	$(TOOLS)/conll_keep_most_frequent_morpho_tags.pl $< $(THRESHOLD) > tmp
-	$(TOOLS)/conllu2mcf -f tmp -1W -2C -3F -4L -5H -6D > $@
+	$(TOOLS)/conllu2mcf -f tmp -1W -2P -3F -4L -5H -6D > $@
 
 test.mcf: $(TEST)
 	$(TOOLS)/conll_keep_most_frequent_morpho_tags.pl $< $(THRESHOLD) > tmp
-	$(TOOLS)/conllu2mcf -f tmp -1W -2C -3F -4L -5H -6D > $@
+	$(TOOLS)/conllu2mcf -f tmp -1W -2P -3F -4L -5H -6D > $@
 
 
 dev.mcf: $(TEST)
 	$(TOOLS)/conll_keep_most_frequent_morpho_tags.pl $< $(THRESHOLD) > tmp
-	$(TOOLS)/conllu2mcf -f tmp -1W -2C -3F -4L -5H -6D > $@
+	$(TOOLS)/conllu2mcf -f tmp -1W -2P -3F -4L -5H -6D > $@
 
 
 clean: