Small improvement of new ranking.

[idzebra-moved-to-github.git] / rset / rsrel.c

diff --git a/rset/rsrel.c b/rset/rsrel.c
index cf945f4..1d3ed61 100644 (file)
--- a/rset/rsrel.c
+++ b/rset/rsrel.c
@@ -1,10 +1,26 @@
 /*
- * 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 1.14  1996-11-08 11:15:58  adam
+ * 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
@@ -54,6 +70,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 #include <assert.h>
 
 #include <isam.h>
@@ -93,6 +110,7 @@ struct rset_rel_info {
     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 */
@@ -152,6 +170,8 @@ static int qcomp (const void *p1, const void *p2)
                             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 */
@@ -162,6 +182,14 @@ static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
     char *isam_tmp_buf;
     int  *isam_r;
     int  *max_tf, *tf;
+
+#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;
@@ -201,6 +229,68 @@ static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
         }
         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;
@@ -258,6 +348,7 @@ static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
         /* 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;
@@ -292,6 +383,7 @@ static void *r_create (const struct rset_control *sel, void *parms,
     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);