diff --git a/maca_morpho/src/fplm.h b/maca_morpho/src/fplm.h
index fe031f15edea7f5aa9257e9e5c32d9ab5ad73ceb..206d1a499bd0f646292670d3c03a2ba833de0d20 100644
--- a/maca_morpho/src/fplm.h
+++ b/maca_morpho/src/fplm.h
@@ -11,6 +11,6 @@
int choose_class(char* class);
int extract_class_position(CLASS class);
void extract_morpho_feature(CLASS class, char* morpho_feature, char* morpho);
- int associate_number_to_classes(char* classes_array, char class);
+ int associate_number_to_classes(FILE* classes_code, int class);
#endif
diff --git a/maca_morpho/src/fplm2cff.c b/maca_morpho/src/fplm2cff.c
index e0321366447c6fcc32e5366ba70ec9b405d15c44..6547410b7c17f9b347ee7bafd1c66baac55a1551 100644
--- a/maca_morpho/src/fplm2cff.c
+++ b/maca_morpho/src/fplm2cff.c
@@ -9,6 +9,7 @@ int main(int argc, char *argv[])
fplm2cff_help_message(ctx);
create_cff(ctx);
printf("cff.txt has been generated in the Files directory.\n");
+ printf("code_class has been generated in the Files directory.\n");
return 0;
}
diff --git a/maca_morpho/src/fplm2cff.h b/maca_morpho/src/fplm2cff.h
index 33b8d98a0718fda0ab6b333dd340ce4643b7b004..ebe8910d911cbd8394e133f9144e06480f5c041c 100644
--- a/maca_morpho/src/fplm2cff.h
+++ b/maca_morpho/src/fplm2cff.h
@@ -10,7 +10,7 @@
#include "fplm.h"
void create_cff(context* ctx);
- void write_cff(FILE *cff, CLASS class, char* classes_array, char* form, char* morpho, feat_vec *fv, feat_model *fm, dico *dico_features);
+ void write_cff(FILE *cff, CLASS class, FILE* code_class, char* form, char* morpho, feat_vec *fv, feat_model *fm, dico *dico_features);
void fplm2cff_help_message(context *ctx);
#endif
diff --git a/maca_morpho/src/fplm2cff_fct.c b/maca_morpho/src/fplm2cff_fct.c
index b2b6e2a1b43837305d0ea06f82186a4dcc1335f5..1f0ceaa96b506f99a67886f38f5a2f67dd361c07 100644
--- a/maca_morpho/src/fplm2cff_fct.c
+++ b/maca_morpho/src/fplm2cff_fct.c
@@ -19,6 +19,7 @@ void create_cff(context* ctx)
{
FILE* fplm = NULL;
FILE* cff = NULL;
+ FILE* code_class = NULL;
feat_vec *fv = NULL;
dico *dico_features = NULL;
feat_model *fm = NULL;
@@ -27,11 +28,7 @@ void create_cff(context* ctx)
char pos[50];
char lemma[100];
char morpho[50];
- char classes_array[100];
CLASS class = choose_class(ctx->class_name);
-
- classes_array[0]='0';
- classes_array[1]='\0';
fplm = fopen(ctx->fplm_filename,"r");
if(fplm == NULL)
{
@@ -47,6 +44,13 @@ void create_cff(context* ctx)
fprintf(stderr,"Problem with the cff file.\n");
exit(EXIT_FAILURE);
}
+ code_class = fopen("../../Files/code_class","w+");
+ if(code_class==NULL)
+ {
+ fprintf(stderr,"Problem with the classes_code file.\n");
+ exit(EXIT_FAILURE);
+ }
+ fprintf(code_class,"%s\n",ctx->class_name);
while((fields_nb = read_line_fplm(fplm, form, pos, lemma, morpho)) != -1)
{
if(fields_nb!=4)
@@ -58,22 +62,23 @@ void create_cff(context* ctx)
}
continue;
}
- write_cff(cff, class, classes_array, form, morpho, fv, fm, dico_features);
+ write_cff(cff, class, code_class, form, morpho, fv, fm, dico_features);
}
if(ctx->features_filename)
dico_print(ctx->features_filename, dico_features);
-
fclose(fplm);
fclose(cff);
+ fclose(code_class);
}
-void write_cff(FILE *cff, CLASS class, char* classes_array, char* form, char* morpho, feat_vec *fv, feat_model *fm, dico *dico_features)
+void write_cff(FILE *cff, CLASS class, FILE* code_class, char* form, char* morpho, feat_vec *fv, feat_model *fm, dico *dico_features)
{
int i=0;
+ int tmp_code;
char morpho_feature[10];
extract_morpho_feature(class, morpho_feature, morpho);
- /*write the class' number in cff (ASCII code of the morpho_feature)*/
+ /*write the class' code in cff*/
if(morpho_feature[0] == '\0')
{
fprintf(cff,"0");
@@ -81,36 +86,29 @@ void write_cff(FILE *cff, CLASS class, char* classes_array, char* form, char* mo
feat_vec_print(cff, fv);
return;
}
- if(class == GENDER || class == NUMBER)
- {
- fprintf(cff, "%d", associate_number_to_classes(classes_array, morpho_feature[i]));
- form2fv(form, fv, fm, dico_features, ADD_MODE);
- feat_vec_print(cff, fv);
- }
- else if(class == PERSON)
- {
- for(i=0; i<(int)strlen(morpho_feature); i++)
- {
- fprintf(cff,"%c", morpho_feature[i]);
- /*because a word can have several persons in morpho so we have to write the
- first class and the features' values, then the second class and the features' values, etc */
- form2fv(form, fv, fm, dico_features, ADD_MODE);
- feat_vec_print(cff, fv);
- }
- }
- else
+ //for example if the tense is 'PS' the class' number will be the ASCII code of 'P' concatenate with the ASCII code of 'S'
+ if(class != PERSON)
{
- //for example if tense is 'PS' the class' number will be the ASCII code of 'P' concatenate with the ASCII code of 'S'
FILE* tmp = fopen("tmp.txt","w+");
for(i=0; i<(int)strlen(morpho_feature); i++)
fprintf(tmp,"%d", morpho_feature[i]);
rewind(tmp);
- fscanf(tmp, "%d", &morpho_feature[0]);
- fprintf(cff,"%d", associate_number_to_classes(classes_array, morpho_feature[0]));
+ fscanf(tmp, "%d", &tmp_code);
+ fprintf(cff,"%d", associate_number_to_classes(code_class,tmp_code));
form2fv(form, fv, fm, dico_features, ADD_MODE);
feat_vec_print(cff, fv);
fclose(tmp);
remove("tmp.txt");
}
+ else
+ {
+ for(i=0; i<(int)strlen(morpho_feature);i++)
+ {
+ tmp_code = morpho_feature[i];
+ fprintf(cff,"%d", associate_number_to_classes(code_class,tmp_code));
+ form2fv(form, fv, fm, dico_features, ADD_MODE);
+ feat_vec_print(cff, fv);
+ }
+ }
}
diff --git a/maca_morpho/src/fplm_fct.c b/maca_morpho/src/fplm_fct.c
index 9b5b050907fdec4453bdad4918a688eb132c91dc..4a40a967d12744c2d8d7130d4eaee5bc4630375b 100644
--- a/maca_morpho/src/fplm_fct.c
+++ b/maca_morpho/src/fplm_fct.c
@@ -3,8 +3,8 @@
#include <string.h>
#include "fplm.h"
-/* Read a line from the fplm file and extract the form/pos/lemma/morpho.
- * Return -1 if there's no more line to read, else the number of string read*/
+/** Read a line from the fplm file and extract the form/pos/lemma/morpho.
+ * Return -1 if there's no more line to read, else the number of string read**/
int read_line_fplm(FILE* fplm, char* form, char* pos, char* lemma, char* morpho)
{
int fields_nb;
@@ -15,8 +15,8 @@ int read_line_fplm(FILE* fplm, char* form, char* pos, char* lemma, char* morpho)
return fields_nb;
}
-/* Return the class' position in morpho
- * (the class could be the tense, the person, the gender or the number of a word)*/
+/** Return the class' position in morpho
+ * (the class could be the tense, the person, the gender or the number of a word)**/
int extract_class_position(CLASS class)
{
switch(class)
@@ -36,7 +36,7 @@ int extract_class_position(CLASS class)
return -1;
}
-/* Return the class choosen by the user if their class exists*/
+/** Return the class choosen by the user if their class exists**/
int choose_class(char* class)
{
if(!strcmp(class,"tense"))
@@ -55,6 +55,7 @@ int choose_class(char* class)
return -1;
}
+/** Extract the class we want from morpho and write it in morpho_feature**/
void extract_morpho_feature(CLASS class, char* morpho_feature, char* morpho)
{
int cpt_diese = 0;
@@ -88,14 +89,18 @@ void extract_morpho_feature(CLASS class, char* morpho_feature, char* morpho)
}
}
-int associate_number_to_classes(char* classes_array, char class)
+/**Write the code_class file (use in predict to know the real class)
+ * Return the class' code**/
+int associate_number_to_classes(FILE* classes_code, int class)
{
- int i;
- int size = (int)strlen(classes_array);
- for(i=1; i<size; i++)
- if(classes_array[i] == class)
- return i;
- classes_array[size] = class;
- classes_array[size+1] = '\0';
- return size;
+ int code=0;
+ int cl;
+ char tmp[10];
+ rewind(classes_code);
+ fscanf(classes_code,"%s",tmp);
+ while(fscanf(classes_code,"%d %d\n",&code,&cl) == 2)
+ if(cl == class)
+ return code;
+ fprintf(classes_code, "%d %d\n", code+1, class);
+ return code+1;
}
diff --git a/maca_morpho/src/maca_morpho_context.c b/maca_morpho/src/maca_morpho_context.c
index 6c32b30a29518fb0326d2bc09fa5dde813631107..c141f8e3d6f28667a79e92ac936ea3014621b246 100644
--- a/maca_morpho/src/maca_morpho_context.c
+++ b/maca_morpho/src/maca_morpho_context.c
@@ -34,6 +34,7 @@ context *context_new(void)
ctx->features_filename = NULL;
ctx->cfw_filename = NULL;
ctx->class_name = NULL;
+ ctx->code_class_name = NULL;
return ctx;
}
@@ -82,6 +83,10 @@ void context_fplm_test_percent_help_message(context *ctx){
fprintf(stderr, "\t-p --percent <int> : percentage of the fplm file to make a test file\n");
}
+void context_code_class_help_message(context* ctx){
+ fprintf(stderr, "\t-y --code_class <file> : code_class file name\n");
+}
+
context *context_read_options(int argc, char *argv[])
{
int c;
@@ -90,7 +95,7 @@ context *context_read_options(int argc, char *argv[])
ctx->program_name = strdup(argv[0]);
- static struct option long_options[12] =
+ static struct option long_options[13] =
{
{"help", no_argument, 0, 'h'},
{"verbose", no_argument, 0, 'v'},
@@ -103,12 +108,13 @@ context *context_read_options(int argc, char *argv[])
{"feat", required_argument, 0, 'x'},
{"weights", required_argument, 0, 'w'},
{"class", required_argument, 0, 'c'},
- {"percent", required_argument, 0, 'p'}
+ {"percent", required_argument, 0, 'p'},
+ {"code_class", required_argument, 0, 'y'}
};
optind = 0;
opterr = 0;
- while ((c = getopt_long (argc, argv, "hvdf:L:M:D:F:x:w:c:p:", long_options, &option_index)) != -1){
+ while ((c = getopt_long (argc, argv, "hvdf:L:M:D:F:y:x:w:c:p:", long_options, &option_index)) != -1){
switch (c)
{
case 'd':
@@ -138,12 +144,15 @@ context *context_read_options(int argc, char *argv[])
case 'w':
ctx->cfw_filename = strdup(optarg);
break;
- case 'c':
- ctx->class_name = strdup(optarg);
- break;
- case 'p':
- ctx->fplm_test_percent = atoi(optarg);
- break;
+ case 'c':
+ ctx->class_name = strdup(optarg);
+ break;
+ case 'p':
+ ctx->fplm_test_percent = atoi(optarg);
+ break;
+ case 'y':
+ ctx->code_class_name = strdup(optarg);
+ break;
}
}
diff --git a/maca_morpho/src/maca_morpho_context.h b/maca_morpho/src/maca_morpho_context.h
index 12443f327b856171e17995aa00c2676761f83613..cf66ad2f4703cc1414ae89afac92b098b5779afe 100644
--- a/maca_morpho/src/maca_morpho_context.h
+++ b/maca_morpho/src/maca_morpho_context.h
@@ -21,6 +21,7 @@ typedef struct {
char *features_filename;
char *cfw_filename;
char *class_name;
+ char *code_class_name;
} context;
@@ -38,4 +39,5 @@ void context_weights_matrix_filename_help_message(context *ctx);
void context_features_model_help_message(context *ctx);
void context_class_help_message(context *ctx);
void context_fplm_test_percent_help_message(context *ctx);
+void context_code_class_help_message(context* ctx);
#endif
diff --git a/maca_morpho/src/predict.h b/maca_morpho/src/predict.h
index 9967e5997970c372d116e3e4b46837d824150988..28dade5b89157489f9eff88cb7a626fef4cb5779 100644
--- a/maca_morpho/src/predict.h
+++ b/maca_morpho/src/predict.h
@@ -11,7 +11,9 @@
#include "fplm.h"
void create_predictions_file(context* ctx);
- void make_prediction(FILE* predictions, char* form, char* morpho, feature_table *cfw, feat_vec *fv, dico *dico_features, feat_model *fm);
+ void make_prediction(FILE* predictions, FILE* code_class, CLASS cl, int* errors, char* form, char* morpho, feature_table *cfw, feat_vec *fv, dico *dico_features, feat_model *fm);
+ void errors_nb(FILE* code_class, CLASS class, int class_predicted, int* errors, char* morpho);
+ int extract_real_class(FILE* code_class, CLASS class, char* morpho_feature, int* real_class);
void predict_help_message(context *ctx);
#endif
diff --git a/maca_morpho/src/predict_fct.c b/maca_morpho/src/predict_fct.c
index 0afe5ad9125fab6b700abfe6b4235e49dcad4f38..13f0df7146cd0635d262a95c87520b0050b7d78e 100644
--- a/maca_morpho/src/predict_fct.c
+++ b/maca_morpho/src/predict_fct.c
@@ -1,5 +1,6 @@
#include <stdlib.h>
#include <stdio.h>
+#include <string.h>
#include "predict.h"
void predict_help_message(context *ctx)
@@ -11,6 +12,7 @@ void predict_help_message(context *ctx)
context_features_filename_help_message(ctx);
context_weights_matrix_filename_help_message(ctx);
context_features_model_help_message(ctx);
+ context_code_class_help_message(ctx);
exit(1);
}
@@ -18,22 +20,38 @@ void create_predictions_file(context* ctx)
{
FILE* fplm_test = NULL;
FILE* predictions = NULL;
+ FILE* code_class = NULL;
feature_table *cfw = NULL;
feat_vec *fv = NULL;
dico *dico_features = NULL;
feat_model *fm = NULL;
+ int line_nb=0;
int fields_nb;
+ int errors = 0;
char form[100];
char pos[50];
char lemma[100];
char morpho[50];
-
+ char class_name[10];
+ CLASS class;
+ code_class = fopen(ctx->code_class_name,"r");
+ if(code_class==NULL)
+ {
+ fprintf(stderr, "Could not the code_class file.\n");
+ exit(EXIT_FAILURE);
+ }
+ if(fscanf(code_class,"%s",class_name)!=1)
+ {
+ fprintf(stderr, "Your code_class file is not conform.\n");
+ exit(EXIT_FAILURE);
+ }
fplm_test = fopen(ctx->fplm_filename,"r");
if(fplm_test == NULL)
{
fprintf(stderr,"Could not open input file.\nYou can generate a fplm_test file with fplm2train_test\nThe fplm_test file will be in the Files directory.\n");
exit(EXIT_FAILURE);
}
+ class = choose_class(class_name);
cfw = feature_table_load(ctx->cfw_filename, ctx->verbose);
fv = feat_vec_new(10);
dico_features = dico_read(ctx->features_filename, 0.5);
@@ -44,7 +62,6 @@ void create_predictions_file(context* ctx)
fprintf(stderr,"Problem with the prediction file.\n");
exit(EXIT_FAILURE);
}
-
while((fields_nb = read_line_fplm(fplm_test, form, pos, lemma, morpho)) != -1)
{
if(fields_nb!=4)
@@ -56,30 +73,76 @@ void create_predictions_file(context* ctx)
}
continue;
}
- make_prediction(predictions, form, morpho, cfw, fv, dico_features, fm);
+ line_nb++;
+ make_prediction(predictions, code_class, class, &errors, form, morpho, cfw, fv, dico_features, fm);
}
-
+ printf("Error rate : %lf %%\n", (float)errors*100/line_nb);
if(ctx->features_filename)
dico_print(ctx->features_filename, dico_features);
-
fclose(fplm_test);
fclose(predictions);
+ fclose(code_class);
}
-void make_prediction(FILE* predictions, char* form, char* morpho, feature_table *cfw, feat_vec *fv, dico *dico_features, feat_model *fm)
+void make_prediction(FILE* predictions, FILE* code_class, CLASS cl, int* errors, char* form, char* morpho, feature_table *cfw, feat_vec *fv, dico *dico_features, feat_model *fm)
{
int class;
float max;
-
fprintf(predictions, "form = %s\n", form);
form2fv(form, fv, fm, dico_features, LOOKUP_MODE);
class = feature_table_argmax(fv, cfw, &max);
feat_vec_print(predictions, fv);
- fprintf(predictions, "class predicted = %d\n", class);
+ fprintf(predictions, "class predicted = %d ", class);
+ errors_nb(code_class, cl, class, errors, morpho);
}
+void errors_nb(FILE* code_class, CLASS class, int class_predicted, int* errors, char* morpho)
+{
+ int i;
+ int size = 0;
+ int real_class[10];
+ char morpho_feature[10];
+ extract_morpho_feature(class,morpho_feature,morpho);
+ size = extract_real_class(code_class, class, morpho_feature, real_class);
+ for(i=0; i<=size; i++)
+ if(class_predicted == real_class[i])
+ return;
+ *errors = *errors+1;
+}
-
+int extract_real_class(FILE* code_class, CLASS class, char* morpho_feature, int* real_class)
+{
+ int size = -1;
+ int tmp_code;
+ int i;
+ if(morpho_feature[0]=='\0')
+ {
+ size++;
+ real_class[size]=0;
+ }
+ else if(class != PERSON)
+ {
+ FILE* tmp = fopen("tmp.txt","w+");
+ for(i=0; i<(int)strlen(morpho_feature); i++)
+ fprintf(tmp,"%d", morpho_feature[i]);
+ rewind(tmp);
+ fscanf(tmp, "%d", &tmp_code);
+ size++;
+ real_class[size] = associate_number_to_classes(code_class,tmp_code);
+ fclose(tmp);
+ remove("tmp.txt");
+ }
+ else
+ {
+ for(i=0; i<(int)strlen(morpho_feature);i++)
+ {
+ size++;
+ tmp_code = morpho_feature[i];
+ real_class[size] = associate_number_to_classes(code_class,tmp_code);
+ }
+ }
+ return size;
+}
diff --git a/perceptron/lib/src/cf_file.c b/perceptron/lib/src/cf_file.c
index 9afcf344fe2a8934f5dd0b71f52d968552a1c0ac..0fecf2ae3bd2a80d70f327229fc2408584062034 100644
--- a/perceptron/lib/src/cf_file.c
+++ b/perceptron/lib/src/cf_file.c
@@ -72,6 +72,7 @@ void look_for_number_of_features_and_classes(char *filename, int *max_feat, int
*max_feat = *max_feat + 1;
*max_class = *max_class + 1;
+ //printf("%d\n",*max_class +1);
fclose(f);
}