diff --git a/maca_trans_parser/src/context.h b/maca_trans_parser/src/context.h
index 9b1b6fed317f1f07e9aab9c762dbd9d1d6e153b1..13cd87663a44951f1e8e3900d4328ae79b2cfb6a 100644
--- a/maca_trans_parser/src/context.h
+++ b/maca_trans_parser/src/context.h
@@ -37,7 +37,10 @@
#define DEFAULT_MODEL_PARSER_NN_FILENAME "maca_trans_parser_nn.weights"
#define DEFAULT_JSON_PARSER_NN_FILENAME "maca_trans_parser_nn.json"
-
+#define DEFAULT_PATH_RELAT "../data/treebank/"
+#define DEFAULT_CONLL07_DEV "dev.conll07"
+#define DEFAULT_CONLL07_TRAIN "train.conll07"
+#define DEFAULT_CONLL07_TEST "test.conll07"
#define DEFAULT_F2P_FILENAME "fP"
#define DEFAULT_FPLM_FILENAME "fplm"
diff --git a/maca_trans_parser/src/maca_error_predictor_tagger.c b/maca_trans_parser/src/maca_error_predictor_tagger.c
index f0ce5148463743607d876639df2a712a6a911e2a..68b4fb6a3903166018d576e5d3c265b1769e9fef 100644
--- a/maca_trans_parser/src/maca_error_predictor_tagger.c
+++ b/maca_trans_parser/src/maca_error_predictor_tagger.c
@@ -54,7 +54,7 @@ void add_signature_to_words_in_word_buffer_tagger(word_buffer *bf, form2pos *f2p
int i;
word *w;
char lower_form[1000];
-
+
for(i = word_buffer_get_nbelem(bf) - 1; i >=0 ; i--){
w = word_buffer_get_word_n(bf, i);
if(word_get_signature(w) != -1) break;
@@ -72,29 +72,24 @@ void add_signature_to_words_in_word_buffer_tagger(word_buffer *bf, form2pos *f2p
void maca_error_predictor_help_message(context *ctx)
{
context_general_help_message(ctx);
- context_mode_help_message(ctx);
+ //context_mode_help_message(ctx);
context_sent_nb_help_message(ctx);
- context_mcd_help_message(ctx);
-
+ //context_mcd_help_message(ctx);
fprintf(stderr, "INPUT\n");
- context_conll_help_message(ctx);
- fprintf(stderr, "IN TEST MODE\n");
+ fprintf(stderr, "\t-i --input <file> : input is in conll07 format (default is dev.conll07)\n");
+ //fprintf(stderr, "IN TEST MODE\n");
context_vocabs_help_message(ctx);
fprintf(stderr, "OUTPUT\n");
- context_cff_help_message(ctx);
- fprintf(stderr, "IN TRAIN MODE\n");
- context_vocabs_help_message(ctx);
+ fprintf(stderr, "\t-x --cff <file> : CFF format file name (default is stdout)\n");
+ //fprintf(stderr, "IN TRAIN MODE\n");
+ //context_vocabs_help_message(ctx);
}
void maca_error_predictor_check_options(context *ctx)
{
- if(0 /*!ctx->input_filename
- || ctx->help
- / || !ctx->mcd_filename /
- || !(ctx->cff_filename || ctx->fann_filename)
- */){
+ if(ctx->help){
maca_error_predictor_help_message(ctx);
exit(1);
}
@@ -106,71 +101,75 @@ int config_is_equal_tagger(config *c1, config *c2)
return ((bm1p(c1)==bm1p(c2))&&(bm2p(c1)==bm2p(c2))&&(bm3p(c1)==bm3p(c2)));
}
+
void generate_error_train(FILE *output_file, context *ctx)
{
config *config_oracle;
- feature_table *ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);
feat_vec *fv_oracle = feat_vec_new(feature_types_nb);
FILE *conll_file_oracle = myfopen(ctx->input_filename, "r");
int postag_oracle;
- float max;
word *b0;
- dico *dico_pos = dico_vec_get_dico(ctx->vocabs, (char *)"POS");
config *config_predicted;
+ feature_table *ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);
feat_vec *fv_predicted = feat_vec_new(feature_types_nb);
FILE *conll_file_predicted = myfopen(ctx->input_filename, "r");
int postag_predicted;
-
+ float max;
+ dico *dico_pos = dico_vec_get_dico(ctx->vocabs, (char *)"POS");
+
config_oracle = config_new(conll_file_oracle, ctx->mcd_struct, 5);
config_predicted = config_new(conll_file_predicted, ctx->mcd_struct, 5);
while(!config_is_terminal(config_oracle)){
- /*if(ctx->f2p){
- add_signature_to_words_in_word_buffer_tagger(config_oracle->bf, ctx->f2p);
+ if(ctx->f2p){
add_signature_to_words_in_word_buffer_tagger(config_predicted->bf, ctx->f2p);
- }*/
+ add_signature_to_words_in_word_buffer_tagger(config_oracle->bf, ctx->f2p);
+ }
// oracle
config2feat_vec_cff(ctx->features_model, config_oracle, ctx->d_perceptron_features, fv_oracle, LOOKUP_MODE);
postag_oracle = oracle_tagger(config_oracle);
- printf("Oracle : ");
- print_word_simple(word_buffer_b0(config_oracle->bf), ctx->mcd_struct, dico_pos, postag_oracle);
+
+ if(ctx->debug_mode){
+ printf("Oracle : ");
+ print_word_simple(word_buffer_b0(config_oracle->bf), ctx->mcd_struct, dico_pos, postag_oracle);
+ }
// predicted
b0 = word_buffer_b0(config_predicted->bf);
config2feat_vec_cff(ctx->features_model, config_predicted, ctx->d_perceptron_features, fv_predicted, LOOKUP_MODE);
postag_predicted = feature_table_argmax(fv_predicted, ft, &max);
-
- printf("Predicted : ");
- print_word_simple(b0, ctx->mcd_struct, dico_pos, postag_predicted);
-
- if(1){
+
+ if(ctx->debug_mode){
+ printf("Predicted : ");
+ print_word_simple(b0, ctx->mcd_struct, dico_pos, postag_predicted);
+ }
+
+ if(ctx->debug_mode){
vcode *vcode_array = feature_table_get_vcode_array(fv_predicted, ft);
for(int i=0; i < 3; i++){
- fprintf(stdout, "%d\t", i);
- fprintf(stdout, "%s\t%.4f\n", dico_int2string(dico_pos, vcode_array[i].class_code), vcode_array[i].score);
+ printf("%d\t", i);
+ printf("%s\t%.4f\n", dico_int2string(dico_pos, vcode_array[i].class_code), vcode_array[i].score);
}
free(vcode_array);
- }
+
+ if (postag_oracle!=postag_predicted)
+ fprintf(stdout, "**************** DIFFERENT CHOICE ***********\n\n");
-
-
- if (postag_oracle!=postag_predicted)
- fprintf(stdout, "**************** DIFFERENT CHOICE ***********\n\n");
+ else
+ fprintf(stdout, "**************** EQUAL CHOICE ***********\n\n");
- else
- fprintf(stdout, "**************** EQUAL CHOICE ***********\n\n");
-
+ }
movement_tagger(config_oracle, postag_oracle);
movement_tagger(config_predicted, postag_predicted);
- fprintf(output_file, "%d", ((config_is_equal_tagger(config_oracle, config_predicted)) ? 1 : 0));
- fprintf(output_file, " or : %d, pred : %d", postag_oracle,postag_predicted);
- feat_vec_print(output_file, fv_predicted);
- //word_set_pos(word_buffer_bm1(config_predicted->bf), postag_oracle);
-
+ if(!ctx->debug_mode || output_file!=stdout) {
+ fprintf(output_file, "%d", ((config_is_equal_tagger(config_oracle, config_predicted)) ? 1 : 0));
+ feat_vec_print(output_file, fv_predicted);
+ }
}
+
feat_vec_free(fv_oracle);
feat_vec_free(fv_predicted);
@@ -183,6 +182,54 @@ void generate_error_train(FILE *output_file, context *ctx)
}
+void error_tagger_set_linguistic_resources_filename(context *ctx)
+{
+ char absolute_filename[500];
+
+ if(!ctx->perc_model_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_MODEL_TAGGER_FILENAME);
+ ctx->perc_model_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->vocabs_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_VOCABS_TAGGER_FILENAME);
+ ctx->vocabs_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->input_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_PATH_RELAT);
+ strcat(absolute_filename, DEFAULT_CONLL07_DEV);
+ ctx->input_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->cff_filename){
+ //printf("cff -> stdout\n")
+ }
+
+ if(!ctx->features_model_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_FEATURES_MODEL_TAGGER_FILENAME);
+ ctx->features_model_filename = strdup(absolute_filename);
+ }
+
+ if(!ctx->f2p_filename){
+ strcpy(absolute_filename, ctx->maca_data_path);
+ strcat(absolute_filename, DEFAULT_F2P_FILENAME);
+ ctx->f2p_filename = strdup(absolute_filename);
+ }
+
+ if(ctx->verbose){
+ fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
+ fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
+ fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
+ fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
+ fprintf(stderr, "f2p_filename = %s\n", ctx->f2p_filename);
+ fprintf(stderr, "input_filename = %s\n", ctx->f2p_filename);
+ }
+}
int main(int argc, char *argv[])
@@ -191,20 +238,13 @@ int main(int argc, char *argv[])
FILE *output_file;
ctx = context_read_options(argc, argv);
- //maca_error_predictor_check_options(ctx);
-
- ctx->perc_model_filename = "/home/mathis/maca_data2/fr/bin/maca_trans_tagger.model" ;
- ctx->features_model_filename = "/home/mathis/maca_data2/fr/bin/maca_trans_tagger.fm" ;
- ctx->input_filename = "/home/mathis/maca_data2/fr/data/treebank/dev.conll07";
- ctx->f2p_filename = "/home/mathis/maca_data2/fr/bin/fP";
- ctx->vocabs_filename = "/home/mathis/maca_data2/fr/bin/maca_trans_tagger.vocab";
- ctx->cff_filename = "/home/mathis/test/stage/error.cff";
-
+ error_tagger_set_linguistic_resources_filename(ctx);
+ ctx->f2p = form2pos_read(ctx->f2p_filename);
+ maca_error_predictor_check_options(ctx);
+
+ /* load ctx */
ctx->mcd_struct = mcd_build_conll07();
-
- //decode_tagger_set_linguistic_resources_filenames(ctx);
- ctx->features_model = feat_model_read(ctx->features_model_filename, feat_lib_build(), 1);//ctx->verbose);
-
+ ctx->features_model = feat_model_read(ctx->features_model_filename, feat_lib_build(), ctx->verbose);
ctx->vocabs = dico_vec_read(ctx->vocabs_filename, ctx->hash_ratio);
mcd_link_to_dico(ctx->mcd_struct, ctx->vocabs, ctx->verbose);
ctx->d_perceptron_features = dico_vec_get_dico(ctx->vocabs, (char *)"d_perceptron_features");
@@ -219,13 +259,13 @@ int main(int argc, char *argv[])
output_file = myfopen(ctx->cff_filename, "w");
else
output_file = stdout;
-
+
generate_error_train(output_file,ctx);
if(ctx->cff_filename)
fclose(output_file);
- //context_free(ctx);
+ context_free(ctx);
return 0;
}