* Sebastian Hammer, Adam Dickmeiss
*
* $Log: rsrel.c,v $
- * Revision 1.8 1995-12-05 11:25:45 adam
+ * Revision 1.12 1996-10-08 13:00:40 adam
+ * Bug fix: result sets with ranked operands in boolean operations weren't
+ * sorted.
+ *
+ * Revision 1.11 1996/10/07 16:05:29 quinn
+ * Work.
+ *
+ * Revision 1.9 1995/12/11 09:15:26 adam
+ * New set types: sand/sor/snot - ranked versions of and/or/not in
+ * ranked/semi-ranked result sets.
+ * Note: the snot not finished yet.
+ * New rset member: flag.
+ * Bug fix: r_delete in rsrel.c did free bad memory block.
+ *
+ * Revision 1.8 1995/12/05 11:25:45 adam
* Doesn't include math.h.
*
* Revision 1.7 1995/10/12 12:41:57 adam
#include <rsrel.h>
#include <alexutil.h>
-static void *r_create(const struct rset_control *sel, void *parms);
+static void *r_create(const struct rset_control *sel, void *parms,
+ int *flags);
static RSFD r_open (RSET ct, int flag);
static void r_close (RSFD rfd);
static void r_delete (RSET ct);
static const rset_control control =
{
- "relevance set type",
+ "relevance",
r_create,
r_open,
r_close,
qsort_info->key_buf + i2*qsort_info->key_size);
}
+#define SCORE_SHOW 0.0 /* base score for showing up */
+#define SCORE_COOC 0.3 /* component dependent on co-oc */
+#define SCORE_DYN (1-(SCORE_SHOW+SCORE_COOC)) /* dynamic component of score */
+
static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
{
char **isam_buf;
char *isam_tmp_buf;
int *isam_r;
- int *max_tf;
+ int *max_tf, *tf;
ISPT *isam_pt;
- double *wgt;
int i;
logf (LOG_DEBUG, "relevance");
isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
isam_tmp_buf = xmalloc (info->key_size);
- max_tf = xmalloc (sizeof (*max_tf) * parms->no_isam_positions);
- wgt = xmalloc (sizeof (*wgt) * parms->no_isam_positions);
+ max_tf = xmalloc (sizeof (*max_tf) * parms->no_terms);
+ tf = xmalloc (sizeof (*tf) * parms->no_terms);
+ for (i = 0; i<parms->no_terms; i++)
+ max_tf[i] = 0;
for (i = 0; i<parms->no_isam_positions; i++)
{
isam_buf[i] = xmalloc (info->key_size);
isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
- max_tf [i] = is_numkeys (isam_pt[i]);
+ max_tf [parms->term_no[i]] = is_numkeys (isam_pt[i]);
isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
}
while (1)
{
- int min = -1, i;
+ int min = -1, i, r;
double score;
+ int co_oc, last_term; /* Number of co-occurrences */
+ last_term = -1;
/* find min with lowest sysno */
for (i = 0; i<parms->no_isam_positions; i++)
+ {
if (isam_r[i] &&
- (min < 0 || (*parms->cmp)(isam_buf[i], isam_buf[min]) < 1))
+ (min < 0 || (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 2))
+ {
min = i;
+ co_oc = 1;
+ }
+ else if (!r && last_term != parms->term_no[i]) /* new occurrence */
+ co_oc++;
+ last_term = parms->term_no[i];
+ }
+
if (min < 0)
break;
memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
/* calculate for all with those sysno */
+ for (i = 0; i < parms->no_terms; i++)
+ tf[i] = 0;
for (i = 0; i<parms->no_isam_positions; i++)
{
int r;
r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
else
r = 2;
- if (r > 1 || r < -1)
- wgt[i] = 0.0;
- else
+ if (r <= 1 && r >= -1)
{
- int tf = 0;
do
{
- tf++;
+ tf[parms->term_no[i]]++;
isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
} while (isam_r[i] &&
(*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
- wgt[i] = 0.1+tf*0.9/max_tf[i];
}
}
/* calculate relevance value */
score = 0.0;
- for (i = 0; i<parms->no_isam_positions; i++)
- score += wgt[i];
+ for (i = 0; i<parms->no_terms; i++)
+ if (tf[i])
+ score += SCORE_SHOW + SCORE_COOC*co_oc/parms->no_terms +
+ SCORE_DYN*tf[i]/max_tf[i];
/* if value is in the top score, then save it - don't emit yet */
- add_rec (info, score, isam_tmp_buf);
+ add_rec (info, score/parms->no_terms, isam_tmp_buf);
}
for (i = 0; i<info->no_rec; i++)
info->sysno_idx[i] = i;
xfree (isam_buf);
xfree (isam_r);
xfree (isam_pt);
- xfree (wgt);
+ xfree(tf);
}
-static void *r_create (const struct rset_control *sel, void *parms)
+static void *r_create (const struct rset_control *sel, void *parms,
+ int *flags)
{
rset_relevance_parms *r_parms = parms;
struct rset_rel_info *info;
+ *flags |= RSET_FLAG_RANKED;
info = xmalloc (sizeof(struct rset_rel_info));
info->key_size = r_parms->key_size;
assert (info->key_size > 1);
xfree (info->sort_idx);
xfree (info->sysno_idx);
xfree (info);
- xfree (ct);
}
static void r_rewind (RSFD rfd)
projects / idzebra-moved-to-github.git / blobdiff