/**
* Print the content of a feature vector
* @param fv feature vector
*/
void fvec_print(fvec_t *fv)
{
assert(fv);
int i, j;
printf("feature vector\n len: %lu, total: %lu, mem: %.2fkb\n",
fv->len, fv->total, fv->mem / 1e3);
if (fv->src)
printf(" src: '%s'\n", fv->src);
if (verbose < 3)
return;
for (i = 0; i < fv->len; i++) {
printf(" 0x%.16llx: %6.4f", (long long unsigned int) fv->dim[i],
fv->val[i]);
/* Lookup feature */
fentry_t *fe = ftable_get(fv->dim[i]);
if (!fe) {
printf("\n");
continue;
}
/* Print feature string */
printf(" [");
for (j = 0; j < fe->len; j++) {
if (isprint(fe->data[j]) || fe->data[j] == '%')
printf("%c", fe->data[j]);
else
printf("%%%.2x", fe->data[j]);
}
printf("]\n");
}
}
/**
* Print shared n-grams for each cluster
* @param c Clustering structure
* @param fa Array of feature vectors
* @param file Output file
*/
void export_shared_ngrams(cluster_t *c, farray_t *fa, const char *file)
{
assert(c && fa && file);
int i, j, k;
double shared;
FILE *f;
char *name = NULL;
config_lookup_float(&cfg, "cluster.shared_ngrams", &shared);
if (shared <= 0.0)
return;
if (verbose > 0)
printf("Exporting shared n-grams with minimum ratio %4.2f.\n",
shared);
if (!(f = fopen(file, "a"))) {
error("Could not create file '%s'.", file);
return;
}
/* Print incremental header */
fprintf(f, "# ---\n# Shared n-grams for %s\n", fa->src);
fprintf(f, "# Minimum ratio of shared n-grams: %4.2f (%2.0f%%)\n",
shared, shared * 100);
fprintf(f, "# ---\n# <cluster> <ratio> <hash> <ngram>\n");
/* Compute shared n-grams per cluster */
for (i = 0; i < c->num; i++) {
fvec_t *s = fvec_zero();
for (j = 0, k = 0; j < c->len; j++) {
if (c->cluster[j] != i)
continue;
/* Clone and binarize */
fvec_t *x = fvec_clone(fa->x[j]);
fvec_bin(x);
if (k == 0)
name = cluster_get_name(c, j);
/* Merge n-grams in cluster */
fvec_t *y = fvec_add(s, x);
fvec_destroy(s);
fvec_destroy(x);
s = y;
k++;
}
/* Check for empty cluster */
if (k == 0)
continue;
fvec_div(s, k);
/* Output shared n-grams */
for (j = 0; j < s->len; j++) {
if (s->val[j] < shared)
continue;
fprintf(f, "%s %6.4f %.16llx ", name, s->val[j],
(long long unsigned int) s->dim[j]);
/* Lookup feature */
fentry_t *fe = ftable_get(s->dim[j]);
if (!fe)
error("Oops. Feature not in lookup table.");
/* Print feature */
fprintf(f, "\"");
for (k = 0; k < fe->len; k++) {
if (isprint(fe->data[k]) || fe->data[k] == '%')
fprintf(f, "%c", fe->data[k]);
else
fprintf(f, "%%%.2x", fe->data[k]);
}
fprintf(f, "\"\n");
}
fvec_destroy(s);
}
fclose(f);
}
