2 * Copyright (C) 1994-1997, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.20 1997-10-31 12:37:55 adam
8 * Code calls xfree() instead of free().
10 * Revision 1.19 1997/10/01 11:44:06 adam
11 * Small improvement of new ranking.
13 * Revision 1.18 1997/09/24 13:36:41 adam
14 * More work on new ranking algorithm.
16 * Revision 1.17 1997/09/22 12:39:07 adam
17 * Added get_pos method for the ranked result sets.
19 * Revision 1.16 1997/09/17 12:19:23 adam
20 * Zebra version corresponds to YAZ version 1.4.
21 * Changed Zebra server so that it doesn't depend on global common_resource.
23 * Revision 1.15 1997/09/09 13:38:16 adam
24 * Partial port to WIN95/NT.
26 * Revision 1.14 1996/11/08 11:15:58 adam
27 * Compressed isam fully supported.
29 * Revision 1.13 1996/10/29 13:55:26 adam
30 * Include of zebrautl.h instead of alexutil.h.
32 * Revision 1.12 1996/10/08 13:00:40 adam
33 * Bug fix: result sets with ranked operands in boolean operations weren't
36 * Revision 1.11 1996/10/07 16:05:29 quinn
39 * Revision 1.9 1995/12/11 09:15:26 adam
40 * New set types: sand/sor/snot - ranked versions of and/or/not in
41 * ranked/semi-ranked result sets.
42 * Note: the snot not finished yet.
43 * New rset member: flag.
44 * Bug fix: r_delete in rsrel.c did free bad memory block.
46 * Revision 1.8 1995/12/05 11:25:45 adam
47 * Doesn't include math.h.
49 * Revision 1.7 1995/10/12 12:41:57 adam
50 * Private info (buf) moved from struct rset_control to struct rset.
51 * Bug fixes in relevance.
53 * Revision 1.6 1995/10/10 14:00:04 adam
54 * Function rset_open changed its wflag parameter to general flags.
56 * Revision 1.5 1995/10/06 14:38:06 adam
57 * New result set method: r_score.
58 * Local no (sysno) and score is transferred to retrieveCtrl.
60 * Revision 1.4 1995/09/14 07:48:56 adam
61 * Other score calculation.
63 * Revision 1.3 1995/09/11 15:23:40 adam
64 * More work on relevance search.
66 * Revision 1.2 1995/09/11 13:09:41 adam
67 * More work on relevance feedback.
69 * Revision 1.1 1995/09/08 14:52:42 adam
70 * Work on relevance feedback.
84 static void *r_create(const struct rset_control *sel, void *parms,
86 static RSFD r_open (RSET ct, int flag);
87 static void r_close (RSFD rfd);
88 static void r_delete (RSET ct);
89 static void r_rewind (RSFD rfd);
90 static int r_count (RSET ct);
91 static int r_read (RSFD rfd, void *buf);
92 static int r_write (RSFD rfd, const void *buf);
93 static int r_score (RSFD rfd, int *score);
95 static const rset_control control =
109 const rset_control *rset_kind_relevance = &control;
111 struct rset_rel_info {
115 int (*cmp)(const void *p1, const void *p2);
116 int (*get_pos)(const void *p);
117 char *key_buf; /* key buffer */
118 float *score_buf; /* score buffer */
119 int *sort_idx; /* score sorted index */
120 int *sysno_idx; /* sysno sorted index (ring buffer) */
121 struct rset_rel_rfd *rfd_list;
124 struct rset_rel_rfd {
128 struct rset_rel_rfd *next;
129 struct rset_rel_info *info;
132 static void add_rec (struct rset_rel_info *info, double score, void *key)
136 for (i = 0; i<info->no_rec; i++)
138 idx = info->sort_idx[i];
139 if (score <= info->score_buf[idx])
142 if (info->no_rec < info->max_rec)
143 { /* there is room for this entry */
144 for (j = info->no_rec; j > i; --j)
145 info->sort_idx[j] = info->sort_idx[j-1];
146 idx = info->sort_idx[j] = info->no_rec;
150 return; /* score too low */
153 idx = info->sort_idx[0]; /* remove this entry */
156 for (j = 0; j < i; ++j) /* make room */
157 info->sort_idx[j] = info->sort_idx[j+1];
158 info->sort_idx[j] = idx; /* allocate sort entry */
160 memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
161 info->score_buf[idx] = score;
165 static struct rset_rel_info *qsort_info;
167 static int qcomp (const void *p1, const void *p2)
172 return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
173 qsort_info->key_buf + i2*qsort_info->key_size);
176 #define NEW_RANKING 1
178 #define SCORE_SHOW 0.0 /* base score for showing up */
179 #define SCORE_COOC 0.3 /* component dependent on co-oc */
180 #define SCORE_DYN (1-(SCORE_SHOW+SCORE_COOC)) /* dynamic component of score */
182 static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
193 char *isam_prev_buf = NULL;
196 ISPT *isam_pt = NULL;
197 ISAMC_PP *isamc_pp = NULL;
200 logf (LOG_DEBUG, "relevance");
201 isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
202 isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
204 isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
206 isamc_pp = xmalloc (sizeof (*isamc_pp) * parms->no_isam_positions);
209 logf (LOG_FATAL, "No isamc or isam in rs_rel");
212 isam_tmp_buf = xmalloc (info->key_size);
213 max_tf = xmalloc (sizeof (*max_tf) * parms->no_terms);
214 tf = xmalloc (sizeof (*tf) * parms->no_terms);
216 for (i = 0; i<parms->no_terms; i++)
218 for (i = 0; i<parms->no_isam_positions; i++)
220 isam_buf[i] = xmalloc (info->key_size);
223 isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
224 max_tf [parms->term_no[i]] = is_numkeys (isam_pt[i]);
225 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
229 isamc_pp[i] = isc_pp_open (parms->isc, parms->isam_positions[i]);
230 max_tf [parms->term_no[i]] = isc_pp_num (isamc_pp[i]);
231 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
233 logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
240 for (i = 0; i<parms->no_isam_positions; i++)
243 (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 1))
247 pos_tf = xmalloc (sizeof(*pos_tf) * parms->no_isam_positions);
248 isam_prev_buf = xmalloc (info->key_size);
249 fact1 = 100000/parms->no_isam_positions;
250 fact2 = 100000/(parms->no_isam_positions*parms->no_isam_positions);
252 no_occur = score_sum = 0;
253 memcpy (isam_prev_buf, isam_buf[min], info->key_size);
254 for (i = 0; i<parms->no_isam_positions; i++)
258 (*parms->cmp)(isam_buf[min], isam_prev_buf) > 1)
260 logf (LOG_LOG, "final occur = %d ratio=%d",
261 no_occur, score_sum / no_occur);
262 add_rec (info, score_sum / (10000.0*no_occur), isam_prev_buf);
265 no_occur = score_sum = 0;
266 memcpy (isam_prev_buf, isam_buf[min], info->key_size);
267 for (i = 0; i<parms->no_isam_positions; i++)
270 pos = (*parms->get_pos)(isam_buf[min]);
271 logf (LOG_LOG, "pos=%d", pos);
272 for (i = 0; i<parms->no_isam_positions; i++)
274 int d = pos - pos_tf[i];
277 if (pos_tf[i] < 0 && i != min)
282 score_sum += fact2 / d;
284 score_sum += fact1 / d;
287 logf (LOG_LOG, "score_sum = %d", score_sum);
290 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
292 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
294 xfree (isam_prev_buf);
301 int co_oc, last_term; /* Number of co-occurrences */
304 /* find min with lowest sysno */
305 for (i = 0; i<parms->no_isam_positions; i++)
308 (min < 0 || (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 2))
313 else if (!r && last_term != parms->term_no[i]) /* new occurrence */
315 last_term = parms->term_no[i];
320 memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
321 /* calculate for all with those sysno */
322 for (i = 0; i < parms->no_terms; i++)
324 for (i = 0; i<parms->no_isam_positions; i++)
329 r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
332 if (r <= 1 && r >= -1)
336 tf[parms->term_no[i]]++;
338 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
340 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
341 } while (isam_r[i] &&
342 (*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
345 /* calculate relevance value */
347 for (i = 0; i<parms->no_terms; i++)
349 score += SCORE_SHOW + SCORE_COOC*co_oc/parms->no_terms +
350 SCORE_DYN*tf[i]/max_tf[i];
351 /* if value is in the top score, then save it - don't emit yet */
352 add_rec (info, score/parms->no_terms, isam_tmp_buf);
355 for (i = 0; i<info->no_rec; i++)
356 info->sysno_idx[i] = i;
358 qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
359 for (i = 0; i<parms->no_isam_positions; i++)
362 is_pt_free (isam_pt[i]);
364 isc_pp_close (isamc_pp[i]);
368 xfree (isam_tmp_buf);
376 static void *r_create (const struct rset_control *sel, void *parms,
379 rset_relevance_parms *r_parms = parms;
380 struct rset_rel_info *info;
382 *flags |= RSET_FLAG_RANKED;
383 info = xmalloc (sizeof(struct rset_rel_info));
384 info->key_size = r_parms->key_size;
385 assert (info->key_size > 1);
386 info->max_rec = r_parms->max_rec;
387 assert (info->max_rec > 1);
388 info->cmp = r_parms->cmp;
389 info->get_pos = r_parms->get_pos;
391 info->key_buf = xmalloc (info->key_size * info->max_rec);
392 info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
393 info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
394 info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
396 info->rfd_list = NULL;
398 relevance (info, r_parms);
402 static RSFD r_open (RSET ct, int flag)
404 struct rset_rel_rfd *rfd;
405 struct rset_rel_info *info = ct->buf;
407 if (flag & RSETF_WRITE)
409 logf (LOG_FATAL, "relevance set type is read-only");
412 rfd = xmalloc (sizeof(*rfd));
414 rfd->next = info->rfd_list;
416 info->rfd_list = rfd;
421 static void r_close (RSFD rfd)
423 struct rset_rel_info *info = ((struct rset_rel_rfd*)rfd)->info;
424 struct rset_rel_rfd **rfdp;
426 for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
429 *rfdp = (*rfdp)->next;
433 logf (LOG_FATAL, "r_close but no rfd match!");
437 static void r_delete (RSET ct)
439 struct rset_rel_info *info = ct->buf;
441 assert (info->rfd_list == NULL);
442 xfree (info->key_buf);
443 xfree (info->score_buf);
444 xfree (info->sort_idx);
445 xfree (info->sysno_idx);
449 static void r_rewind (RSFD rfd)
451 struct rset_rel_rfd *p = rfd;
452 struct rset_rel_info *info = p->info;
454 if (p->flag & RSETF_SORT_RANK)
455 p->position = info->no_rec;
460 static int r_count (RSET ct)
462 struct rset_rel_info *info = ct->buf;
467 static int r_read (RSFD rfd, void *buf)
469 struct rset_rel_rfd *p = rfd;
470 struct rset_rel_info *info = p->info;
472 if (p->flag & RSETF_SORT_RANK)
474 if (p->position <= 0)
477 p->last_read_pos = info->sort_idx[p->position];
481 if (p->position == info->no_rec)
483 p->last_read_pos = info->sysno_idx[p->position];
487 info->key_buf + info->key_size * p->last_read_pos,
492 static int r_score (RSFD rfd, int *score)
494 struct rset_rel_rfd *p = rfd;
495 struct rset_rel_info *info = p->info;
497 *score = (int) (1000*info->score_buf[p->last_read_pos]);
501 static int r_write (RSFD rfd, const void *buf)
503 logf (LOG_FATAL, "relevance set type is read-only");