/*
- * Copyright (C) 1994-1995, Index Data I/S
+ * Copyright (C) 1994-1997, Index Data I/S
* All rights reserved.
* Sebastian Hammer, Adam Dickmeiss
*
* $Log: rsrel.c,v $
- * Revision 1.11 1996-10-07 16:05:29 quinn
+ * Revision 1.19 1997-10-01 11:44:06 adam
+ * Small improvement of new ranking.
+ *
+ * Revision 1.18 1997/09/24 13:36:41 adam
+ * More work on new ranking algorithm.
+ *
+ * Revision 1.17 1997/09/22 12:39:07 adam
+ * Added get_pos method for the ranked result sets.
+ *
+ * Revision 1.16 1997/09/17 12:19:23 adam
+ * Zebra version corresponds to YAZ version 1.4.
+ * Changed Zebra server so that it doesn't depend on global common_resource.
+ *
+ * Revision 1.15 1997/09/09 13:38:16 adam
+ * Partial port to WIN95/NT.
+ *
+ * Revision 1.14 1996/11/08 11:15:58 adam
+ * Compressed isam fully supported.
+ *
+ * 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
#include <stdio.h>
#include <stdlib.h>
+#include <string.h>
#include <assert.h>
#include <isam.h>
+#include <isamc.h>
#include <rsrel.h>
-#include <alexutil.h>
+#include <zebrautl.h>
static void *r_create(const struct rset_control *sel, void *parms,
int *flags);
int max_rec;
int no_rec;
int (*cmp)(const void *p1, const void *p2);
+ int (*get_pos)(const void *p);
char *key_buf; /* key buffer */
float *score_buf; /* score buffer */
int *sort_idx; /* score sorted index */
qsort_info->key_buf + i2*qsort_info->key_size);
}
+#define NEW_RANKING 1
+
#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 */
char *isam_tmp_buf;
int *isam_r;
int *max_tf, *tf;
- ISPT *isam_pt;
+
+#if NEW_RANKING
+ int *pos_tf = NULL;
+ int score_sum = 0;
+ int no_occur = 0;
+ char *isam_prev_buf = NULL;
+ int fact1, fact2;
+#endif
+ ISPT *isam_pt = NULL;
+ ISAMC_PP *isamc_pp = NULL;
int i;
logf (LOG_DEBUG, "relevance");
isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
- isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
+ if (parms->is)
+ isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
+ else if (parms->isc)
+ isamc_pp = xmalloc (sizeof (*isamc_pp) * parms->no_isam_positions);
+ else
+ {
+ logf (LOG_FATAL, "No isamc or isam in rs_rel");
+ abort ();
+ }
isam_tmp_buf = xmalloc (info->key_size);
max_tf = xmalloc (sizeof (*max_tf) * parms->no_terms);
tf = xmalloc (sizeof (*tf) * parms->no_terms);
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 [parms->term_no[i]] = is_numkeys (isam_pt[i]);
- isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+ if (isam_pt)
+ {
+ isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
+ max_tf [parms->term_no[i]] = is_numkeys (isam_pt[i]);
+ isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+ }
+ else if (isamc_pp)
+ {
+ isamc_pp[i] = isc_pp_open (parms->isc, parms->isam_positions[i]);
+ max_tf [parms->term_no[i]] = isc_pp_num (isamc_pp[i]);
+ isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
+ }
logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
}
+#if NEW_RANKING
+ while (1)
+ {
+ int r, min = -1;
+ int pos = 0;
+ for (i = 0; i<parms->no_isam_positions; i++)
+ if (isam_r[i] &&
+ (min < 0 ||
+ (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 1))
+ min = i;
+ if (!isam_prev_buf)
+ {
+ pos_tf = xmalloc (sizeof(*pos_tf) * parms->no_isam_positions);
+ isam_prev_buf = xmalloc (info->key_size);
+ fact1 = 100000/parms->no_isam_positions;
+ fact2 = 100000/(parms->no_isam_positions*parms->no_isam_positions);
+
+ no_occur = score_sum = 0;
+ memcpy (isam_prev_buf, isam_buf[min], info->key_size);
+ for (i = 0; i<parms->no_isam_positions; i++)
+ pos_tf[i] = -10;
+ }
+ else if (min < 0 ||
+ (*parms->cmp)(isam_buf[min], isam_prev_buf) > 1)
+ {
+ logf (LOG_LOG, "final occur = %d ratio=%d",
+ no_occur, score_sum / no_occur);
+ add_rec (info, score_sum / (10000.0*no_occur), isam_prev_buf);
+ if (min < 0)
+ break;
+ no_occur = score_sum = 0;
+ memcpy (isam_prev_buf, isam_buf[min], info->key_size);
+ for (i = 0; i<parms->no_isam_positions; i++)
+ pos_tf[i] = -10;
+ }
+ pos = (*parms->get_pos)(isam_buf[min]);
+ logf (LOG_LOG, "pos=%d", pos);
+ for (i = 0; i<parms->no_isam_positions; i++)
+ {
+ int d = pos - pos_tf[i];
+
+ no_occur++;
+ if (pos_tf[i] < 0 && i != min)
+ continue;
+ if (d < 10)
+ d = 10;
+ if (i == min)
+ score_sum += fact2 / d;
+ else
+ score_sum += fact1 / d;
+ }
+ pos_tf[min] = pos;
+ logf (LOG_LOG, "score_sum = %d", score_sum);
+ i = min;
+ if (isam_pt)
+ isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+ else if (isamc_pp)
+ isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
+ }
+ xfree (isam_prev_buf);
+ xfree (pos_tf);
+#else
while (1)
{
int min = -1, i, r;
co_oc = 1;
}
else if (!r && last_term != parms->term_no[i]) /* new occurrence */
- co_oc++;
+ co_oc++;
last_term = parms->term_no[i];
}
-
+
if (min < 0)
break;
memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
else
r = 2;
-#if 0
- if (r > 1 || r < -1)
- wgt[parms->term_no[i]] = 0.0;
-#endif
if (r <= 1 && r >= -1)
{
do
{
tf[parms->term_no[i]]++;
- isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+ if (isam_pt)
+ isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
+ else if (isamc_pp)
+ isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
} while (isam_r[i] &&
(*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
}
/* if value is in the top score, then save it - don't emit yet */
add_rec (info, score/parms->no_terms, isam_tmp_buf);
}
+#endif
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]);
+ if (isam_pt)
+ is_pt_free (isam_pt[i]);
+ if (isamc_pp)
+ isc_pp_close (isamc_pp[i]);
xfree (isam_buf[i]);
}
xfree (max_tf);
xfree (isam_buf);
xfree (isam_r);
xfree (isam_pt);
+ xfree (isamc_pp);
xfree(tf);
}
info->max_rec = r_parms->max_rec;
assert (info->max_rec > 1);
info->cmp = r_parms->cmp;
+ info->get_pos = r_parms->get_pos;
info->key_buf = xmalloc (info->key_size * info->max_rec);
info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
projects / idzebra-moved-to-github.git / blobdiff