-/* $Id: ranksimilarity.c,v 1.2 2006-05-03 13:55:20 marc Exp $
- Copyright (C) 1995-2005
+/* $Id: ranksimilarity.c,v 1.12 2007-01-15 15:10:16 adam Exp $
+ Copyright (C) 1995-2007
Index Data ApS
This file is part of the Zebra server.
for more details.
You should have received a copy of the GNU General Public License
-along with Zebra; see the file LICENSE.zebra. If not, write to the
-Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
-02111-1307, USA.
+along with this program; if not, write to the Free Software
+Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+
*/
#include <stdio.h>
int dummy;
};
+/** term specific info and statistics to be used under ranking */
struct ranksimilarity_term_info {
- /* frequency of term within document */
- int freq_term_doc;
- /* frequency of term within result set */
+
+ /** frequency of term within document field */
+ int freq_term_docfield;
+
+ /** frequency of term within result set of given term */
zint freq_term_resset;
- /* rank flag is one if term is to be included in ranking */
+
+ /** number of docs within result set */
+ zint no_docs_resset;
+
+ /** number of docs with this fieldindex in database */
+ zint no_docs_fieldindex;
+
+ /** number of terms in this fieldindex */
+ zint no_terms_fieldindex;
+
+ /** rank flag is one if term is to be included in ranking */
int rank_flag;
- /* relative ranking weight of term */
- int term_weight;
+
+ /** relative ranking weight of term fieldindex */
+ int fieldindex_weight;
+
+ /** term id used to access term name and other info */
TERMID term;
+
+ /** index number in terms[i] array */
int term_index;
};
struct ranksimilarity_set_info {
int last_pos;
- int no_entries;
- int no_rank_entries;
+
+ /** number of terms in query */
+ int no_terms_query;
+
+ /** number of terms in query which are included in ranking */
+ int no_ranked_terms_query;
+
+ /** number of documents in entire database */
+ zint no_docs_database;
+
+ /** number of terms in entire database */
+ zint no_terms_database;
+
+ /** array of size no_terms_query with statistics gathered per term */
struct ranksimilarity_term_info *entries;
+
NMEM nmem;
};
{
int i;
- for (i = 0; i < si->no_entries; i++){
- si->entries[i].freq_term_doc = 0;
- }
+ for (i = 0; i < si->no_terms_query; i++)
+ {
+ si->entries[i].freq_term_docfield = 0;
+ }
}
yaz_log(log_level, "begin() numterms=%d", numterms);
- si->no_entries = numterms;
- /* count how many terms are ranked (2=102 or similar) */
- si->no_rank_entries = 0;
+ /* setting database global statistics */
+ si->no_docs_database = -1; /* TODO */
+ si->no_terms_database = -1; /* TODO */
+
+ /* setting query statistics */
+ si->no_terms_query = numterms;
+ si->no_ranked_terms_query = 0;
+
+ /* setting internal data structures */
si->nmem=nmem;
si->entries = (struct ranksimilarity_term_info *)
nmem_malloc (si->nmem, sizeof(*si->entries)*numterms);
ranksimilar_rec_reset(si);
- /* looping all terms in a specific field of query */
+ /* looping all terms in a specific fieldindex of query */
for (i = 0; i < numterms; i++)
{
struct ord_list *ol = NULL;
{
const char *cp = strstr(terms[i]->flags+4, ",w=");
- (si->no_rank_entries)++;
+ zint no_docs_fieldindex = 0;
+ zint no_terms_fieldindex = 0;
+
+ yaz_log(log_level, "begin() terms[%d]: '%s' flags=%s",
+ i, terms[i]->name, terms[i]->flags);
+
+ (si->no_ranked_terms_query)++;
ol = terms[i]->ol;
+
si->entries[i].rank_flag = 1;
+ /* notice that the call to rset_count(rset) has he side-effect
+ of setting rset->hits_limit = rset_count(rset) ??? */
si->entries[i].freq_term_resset = rset_count(terms[i]->rset);
+ si->entries[i].no_docs_resset = terms[i]->rset->hits_count;
+
+
if (cp)
- si->entries[i].term_weight = atoi (cp+3);
+ si->entries[i].fieldindex_weight = atoi (cp+3);
else
- si->entries[i].term_weight = 34; /* sqrroot of 1000 */
-
- yaz_log(log_level, "begin() terms[%d]: '%s' flags=%s",
- i, terms[i]->name, terms[i]->flags);
+ si->entries[i].fieldindex_weight = 34; /* sqrroot of 1000 */
+
+
+ /*
+ yaz_log(log_level, "begin() rset_count(terms[%d]->rset) = "
+ ZINT_FORMAT, i, rset_count(terms[i]->rset));
+ yaz_log(log_level, "begin() terms[%d]->rset->hits_limit = "
+ ZINT_FORMAT, i, terms[i]->rset->hits_limit);
+ yaz_log(log_level, "begin() terms[%d]->rset->hits_count = "
+ ZINT_FORMAT, i, terms[i]->rset->hits_count);
+ yaz_log(log_level, "begin() terms[%d]->rset->hits_round = "
+ ZINT_FORMAT, i, terms[i]->rset->hits_round);
+ yaz_log(log_level, "begin() terms[%d]->rset->hits_approx = %d",
+ i, terms[i]->rset->hits_approx);
+ */
/* looping indexes where term terms[i] is found */
- for (; ol; ol = ol->next)
+
+ for (; ol; ol = ol->next)
{
int index_type = 0;
const char *db = 0;
const char *string_index = 0;
- int set = -1;
- int use = -1;
-
+
zebraExplain_lookup_ord(reg->zei,
- ol->ord, &index_type, &db, &set, &use,
+ ol->ord, &index_type, &db,
&string_index);
+ no_docs_fieldindex
+ += zebraExplain_ord_get_doc_occurrences(reg->zei, ol->ord);
+ no_terms_fieldindex
+ += zebraExplain_ord_get_term_occurrences(reg->zei, ol->ord);
+
if (string_index)
yaz_log(log_level,
- "begin() index: ord=%d type=%c db=%s str-index=%s",
+ "begin() index: ord=%d type=%c db=%s str-index=%s",
ol->ord, index_type, db, string_index);
else
yaz_log(log_level,
- "begin() index: ord=%d type=%c db=%s set=%d use=%d",
- ol->ord, index_type, db, set, use);
+ "begin() index: ord=%d type=%c db=%s",
+ ol->ord, index_type, db);
}
+ si->entries[i].no_docs_fieldindex = no_docs_fieldindex;
+ si->entries[i].no_terms_fieldindex = no_terms_fieldindex;
}
si->entries[i].term = terms[i];
}
+
/**
* add: Called for each word occurence in a result set. This routine
* should be as fast as possible. This routine should "incrementally"
assert(si);
if (!term)
{
- yaz_log(log_level, "add() seqno=%d NULL term", seqno);
+ /* yaz_log(log_level, "add() seqno=%d NULL term", seqno); */
return;
}
ti= (struct ranksimilarity_term_info *) term->rankpriv;
assert(ti);
si->last_pos = seqno;
- ti->freq_term_doc++;
- yaz_log(log_level, "add() seqno=%d term=%s freq_term_doc=%d",
- seqno, term->name, ti->freq_term_doc);
+ ti->freq_term_docfield++;
+ /*yaz_log(log_level, "add() seqno=%d term=%s freq_term_docfield=%d",
+ seqno, term->name, ti->freq_term_docfield); */
}
/*
= (struct ranksimilarity_set_info *) set_handle;
- yaz_log(log_level, "calc() sysno = %d", sysno);
- yaz_log(log_level, "calc() staticrank = %d", staticrank);
+ yaz_log(log_level, "calc() sysno = " ZINT_FORMAT, sysno);
+ yaz_log(log_level, "calc() staticrank = " ZINT_FORMAT, staticrank);
- yaz_log(log_level, "calc() si->no_entries = %d",
- si->no_entries);
- yaz_log(log_level, "calc() si->no_rank_entries = %d",
- si->no_rank_entries);
+ yaz_log(log_level, "calc() si->no_terms_query = %d",
+ si->no_terms_query);
+ yaz_log(log_level, "calc() si->no_ranked_terms_query = %d",
+ si->no_ranked_terms_query);
+ yaz_log(log_level, "calc() si->no_docs_database = " ZINT_FORMAT,
+ si->no_docs_database);
+ yaz_log(log_level, "calc() si->no_terms_database = " ZINT_FORMAT,
+ si->no_terms_database);
- if (!si->no_rank_entries)
+ if (!si->no_ranked_terms_query)
return -1; /* ranking not enabled for any terms */
/* here goes your formula to compute a scoring function */
/* you may use all the gathered statistics here */
- for (i = 0; i < si->no_entries; i++)
+ for (i = 0; i < si->no_terms_query; i++)
{
- yaz_log(log_level, "calc() entries[%d] termid %d",
+ yaz_log(log_level, "calc() entries[%d] termid %p",
i, si->entries[i].term);
if (si->entries[i].term){
yaz_log(log_level, "calc() entries[%d] term '%s' flags=%s",
i, si->entries[i].term->name, si->entries[i].term->flags);
yaz_log(log_level, "calc() entries[%d] rank_flag %d",
i, si->entries[i].rank_flag );
- yaz_log(log_level, "calc() entries[%d] term_weight %d",
- i, si->entries[i].term_weight );
- yaz_log(log_level, "calc() entries[%d] freq_term_doc %d",
- i, si->entries[i].freq_term_doc );
- yaz_log(log_level, "calc() entries[%d] freq_term_resset %d",
+ yaz_log(log_level, "calc() entries[%d] fieldindex_weight %d",
+ i, si->entries[i].fieldindex_weight );
+ yaz_log(log_level, "calc() entries[%d] freq_term_docfield %d",
+ i, si->entries[i].freq_term_docfield );
+ yaz_log(log_level, "calc() entries[%d] freq_term_resset " ZINT_FORMAT,
i, si->entries[i].freq_term_resset );
+ yaz_log(log_level, "calc() entries[%d] no_docs_resset " ZINT_FORMAT,
+ i, si->entries[i].no_docs_resset );
+ yaz_log(log_level, "calc() entries[%d] no_docs_fieldindex "
+ ZINT_FORMAT,
+ i, si->entries[i].no_docs_fieldindex );
+ yaz_log(log_level, "calc() entries[%d] no_terms_fieldindex "
+ ZINT_FORMAT,
+ i, si->entries[i].no_terms_fieldindex );
}
}
/* staticrank = 0 is highest, MAXINT lowest */
- score = INT_MAX - staticrank; /* but score is reverse (logical) */
+ if (staticrank >= INT_MAX)
+ score = 0;
+ else
+ { /* but score is reverse (logical) */
+ score = INT_MAX - CAST_ZINT_TO_INT(staticrank);
+ }
/* debugging statistics output */
};
struct rank_control *rank_similarity_class = &rank_control;
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+