int vec_len;
struct word_entry *entries;
pp2_charset_token_t prt;
+ int rank_cluster;
NMEM nmem;
};
{
for (; entries; entries = entries->next)
{
- if (!strcmp(norm_str, entries->norm_str))
+ if (*norm_str && !strcmp(norm_str, entries->norm_str))
{
const char *cp = 0;
int no_read = 0;
for (i = 0; i < numwords; i++)
{
const char *norm_str;
-
+
ccl_field = nmem_strdup_null(res->nmem, n->u.t.qual);
pp2_charset_token_first(res->prt, words[i], 0);
}
struct relevance *relevance_create_ccl(pp2_charset_fact_t pft,
- struct ccl_rpn_node *query)
+ struct ccl_rpn_node *query,
+ int rank_cluster)
{
NMEM nmem = nmem_create();
struct relevance *res = nmem_malloc(nmem, sizeof(*res));
res->nmem = nmem;
res->entries = 0;
res->vec_len = 1;
+ res->rank_cluster = rank_cluster;
res->prt = pp2_charset_token_create(pft, "relevance");
-
+
pull_terms(res, query);
res->doc_frequency_vec = nmem_malloc(nmem, res->vec_len * sizeof(int));
for (i = 0; i < res->vec_len; i++)
- res->doc_frequency_vec[i] = 0;
+ res->doc_frequency_vec[i] = 0;
return res;
}
r->vec_len * sizeof(*rec->term_frequency_vec));
for (i = 0; i < r->vec_len; i++)
rec->term_frequency_vec[i] = 0;
-
+
// term frequency divided by length of field [1,...]
rec->term_frequency_vecf =
nmem_malloc(r->nmem,
r->vec_len * sizeof(*rec->term_frequency_vecf));
for (i = 0; i < r->vec_len; i++)
rec->term_frequency_vecf[i] = 0.0;
-
+
// for relevance_countwords (so we don't have to xmalloc/xfree)
rec->term_frequency_vec_tmp =
nmem_malloc(r->nmem,
idfvec[i] = 0;
else
{
- // This conditional may be terribly wrong
- // It was there to address the situation where vec[0] == vec[i]
- // which leads to idfvec[i] == 0... not sure about this
- // Traditional TF-IDF may assume that a word that occurs in every
- // record is irrelevant, but this is actually something we will
- // see a lot
- if ((idfvec[i] = log((float) rel->doc_frequency_vec[0] /
- rel->doc_frequency_vec[i])) < 0.0000001)
- idfvec[i] = 1;
+ /* add one to nominator idf(t,D) to ensure a value > 0 */
+ idfvec[i] = log((float) (1 + rel->doc_frequency_vec[0]) /
+ rel->doc_frequency_vec[i]);
}
}
// Calculate relevance for each document
for (t = 1; t < rel->vec_len; t++)
{
float termfreq = (float) rec->term_frequency_vecf[t];
- relevance += 100000 * (termfreq * idfvec[t] + 0.0000005);
+ relevance += 100000 * termfreq * idfvec[t];
+ }
+ if (!rel->rank_cluster)
+ {
+ struct record *record;
+ int cluster_size = 0;
+
+ for (record = rec->records; record; record = record->next)
+ cluster_size++;
+
+ relevance /= cluster_size;
}
rec->relevance_score = relevance;
}