* Sebastian Hammer, Adam Dickmeiss
*
* $Log: rsrel.c,v $
- * Revision 1.6 1995-10-10 14:00:04 adam
+ * Revision 1.13 1996-10-29 13:55:26 adam
+ * Include of zebrautl.h instead of alexutil.h.
+ *
+ * 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
+ * Private info (buf) moved from struct rset_control to struct rset.
+ * Bug fixes in relevance.
+ *
+ * Revision 1.6 1995/10/10 14:00:04 adam
* Function rset_open changed its wflag parameter to general flags.
*
* Revision 1.5 1995/10/06 14:38:06 adam
#include <stdio.h>
#include <stdlib.h>
-#include <math.h>
#include <assert.h>
#include <isam.h>
#include <rsrel.h>
-#include <alexutil.h>
+#include <zebrautl.h>
-static rset_control *r_create(const struct rset_control *sel, void *parms);
-static RSFD r_open (rset_control *ct, int flag);
+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_control *ct);
+static void r_delete (RSET ct);
static void r_rewind (RSFD rfd);
-static int r_count (rset_control *ct);
+static int r_count (RSET ct);
static int r_read (RSFD rfd, void *buf);
static int r_write (RSFD rfd, const void *buf);
static int r_score (RSFD rfd, int *score);
static const rset_control control =
{
- "relevance set type",
- 0,
+ "relevance",
r_create,
r_open,
r_close,
char *key_buf; /* key buffer */
float *score_buf; /* score buffer */
int *sort_idx; /* score sorted index */
- int *sysno_idx; /* sysno sorted index (ring buffer) */
- int sysno_idx_p; /* last sysno sort index */
+ int *sysno_idx; /* sysno sorted index (ring buffer) */
struct rset_rel_rfd *rfd_list;
};
struct rset_rel_rfd {
+ int last_read_pos;
int position;
int flag;
struct rset_rel_rfd *next;
{
int idx, i, j;
- logf (LOG_DEBUG, "add %f", score);
for (i = 0; i<info->no_rec; i++)
{
idx = info->sort_idx[i];
info->sort_idx[j] = info->sort_idx[j+1];
info->sort_idx[j] = idx; /* allocate sort entry */
}
- info->sysno_idx[info->sysno_idx_p] = idx;
- if (++(info->sysno_idx_p) == info->max_rec)
- info->sysno_idx_p = 0;
-
memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
info->score_buf[idx] = score;
}
+
+static struct rset_rel_info *qsort_info;
+
+static int qcomp (const void *p1, const void *p2)
+{
+ int i1 = *(int*) p1;
+ int i2 = *(int*) p2;
+
+ return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
+ 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;
+ qsort_info = info;
+ qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
for (i = 0; i<parms->no_isam_positions; i++)
{
is_pt_free (isam_pt[i]);
xfree (isam_buf);
xfree (isam_r);
xfree (isam_pt);
- xfree (wgt);
+ xfree(tf);
}
-static rset_control *r_create (const struct rset_control *sel, void *parms)
+static void *r_create (const struct rset_control *sel, void *parms,
+ int *flags)
{
- rset_control *newct;
rset_relevance_parms *r_parms = parms;
struct rset_rel_info *info;
- newct = xmalloc(sizeof(*newct));
- memcpy(newct, sel, sizeof(*sel));
- newct->buf = xmalloc (sizeof(struct rset_rel_info));
-
- info = newct->buf;
+ *flags |= RSET_FLAG_RANKED;
+ info = xmalloc (sizeof(struct rset_rel_info));
info->key_size = r_parms->key_size;
assert (info->key_size > 1);
info->max_rec = r_parms->max_rec;
info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
- info->sysno_idx_p = 0;
info->no_rec = 0;
info->rfd_list = NULL;
relevance (info, r_parms);
- return newct;
+ return info;
}
-static RSFD r_open (rset_control *ct, int flag)
+static RSFD r_open (RSET ct, int flag)
{
struct rset_rel_rfd *rfd;
struct rset_rel_info *info = ct->buf;
rfd = xmalloc (sizeof(*rfd));
rfd->flag = flag;
rfd->next = info->rfd_list;
- info->rfd_list = rfd;
- rfd->position = info->no_rec;
rfd->info = info;
+ info->rfd_list = rfd;
+ r_rewind (rfd);
return rfd;
}
assert (0);
}
-static void r_delete (rset_control *ct)
+static void r_delete (RSET ct)
{
struct rset_rel_info *info = ct->buf;
xfree (info->sort_idx);
xfree (info->sysno_idx);
xfree (info);
- xfree (ct);
}
static void r_rewind (RSFD rfd)
{
struct rset_rel_rfd *p = rfd;
struct rset_rel_info *info = p->info;
-
- p->position = info->no_rec;
+
+ if (p->flag & RSETF_SORT_RANK)
+ p->position = info->no_rec;
+ else
+ p->position = 0;
}
-static int r_count (rset_control *ct)
+static int r_count (RSET ct)
{
struct rset_rel_info *info = ct->buf;
struct rset_rel_rfd *p = rfd;
struct rset_rel_info *info = p->info;
- if (p->position <= 0)
- return 0;
- --(p->position);
+ if (p->flag & RSETF_SORT_RANK)
+ {
+ if (p->position <= 0)
+ return 0;
+ --(p->position);
+ p->last_read_pos = info->sort_idx[p->position];
+ }
+ else
+ {
+ if (p->position == info->no_rec)
+ return 0;
+ p->last_read_pos = info->sysno_idx[p->position];
+ ++(p->position);
+ }
memcpy ((char*) buf,
- info->key_buf + info->key_size * info->sort_idx[p->position],
+ info->key_buf + info->key_size * p->last_read_pos,
info->key_size);
return 1;
}
struct rset_rel_rfd *p = rfd;
struct rset_rel_info *info = p->info;
- if (p->position < 0)
- return 0;
- *score = (int) (1000*info->score_buf[info->sort_idx[p->position]]);
+ *score = (int) (1000*info->score_buf[p->last_read_pos]);
return 1;
}