2 * Copyright (C) 1994-1997, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.17 1997-09-22 12:39:07 adam
8 * Added get_pos method for the ranked result sets.
10 * Revision 1.16 1997/09/17 12:19:23 adam
11 * Zebra version corresponds to YAZ version 1.4.
12 * Changed Zebra server so that it doesn't depend on global common_resource.
14 * Revision 1.15 1997/09/09 13:38:16 adam
15 * Partial port to WIN95/NT.
17 * Revision 1.14 1996/11/08 11:15:58 adam
18 * Compressed isam fully supported.
20 * Revision 1.13 1996/10/29 13:55:26 adam
21 * Include of zebrautl.h instead of alexutil.h.
23 * Revision 1.12 1996/10/08 13:00:40 adam
24 * Bug fix: result sets with ranked operands in boolean operations weren't
27 * Revision 1.11 1996/10/07 16:05:29 quinn
30 * Revision 1.9 1995/12/11 09:15:26 adam
31 * New set types: sand/sor/snot - ranked versions of and/or/not in
32 * ranked/semi-ranked result sets.
33 * Note: the snot not finished yet.
34 * New rset member: flag.
35 * Bug fix: r_delete in rsrel.c did free bad memory block.
37 * Revision 1.8 1995/12/05 11:25:45 adam
38 * Doesn't include math.h.
40 * Revision 1.7 1995/10/12 12:41:57 adam
41 * Private info (buf) moved from struct rset_control to struct rset.
42 * Bug fixes in relevance.
44 * Revision 1.6 1995/10/10 14:00:04 adam
45 * Function rset_open changed its wflag parameter to general flags.
47 * Revision 1.5 1995/10/06 14:38:06 adam
48 * New result set method: r_score.
49 * Local no (sysno) and score is transferred to retrieveCtrl.
51 * Revision 1.4 1995/09/14 07:48:56 adam
52 * Other score calculation.
54 * Revision 1.3 1995/09/11 15:23:40 adam
55 * More work on relevance search.
57 * Revision 1.2 1995/09/11 13:09:41 adam
58 * More work on relevance feedback.
60 * Revision 1.1 1995/09/08 14:52:42 adam
61 * Work on relevance feedback.
75 static void *r_create(const struct rset_control *sel, void *parms,
77 static RSFD r_open (RSET ct, int flag);
78 static void r_close (RSFD rfd);
79 static void r_delete (RSET ct);
80 static void r_rewind (RSFD rfd);
81 static int r_count (RSET ct);
82 static int r_read (RSFD rfd, void *buf);
83 static int r_write (RSFD rfd, const void *buf);
84 static int r_score (RSFD rfd, int *score);
86 static const rset_control control =
100 const rset_control *rset_kind_relevance = &control;
102 struct rset_rel_info {
106 int (*cmp)(const void *p1, const void *p2);
107 int (*get_pos)(const void *p);
108 char *key_buf; /* key buffer */
109 float *score_buf; /* score buffer */
110 int *sort_idx; /* score sorted index */
111 int *sysno_idx; /* sysno sorted index (ring buffer) */
112 struct rset_rel_rfd *rfd_list;
115 struct rset_rel_rfd {
119 struct rset_rel_rfd *next;
120 struct rset_rel_info *info;
123 static void add_rec (struct rset_rel_info *info, double score, void *key)
127 for (i = 0; i<info->no_rec; i++)
129 idx = info->sort_idx[i];
130 if (score <= info->score_buf[idx])
133 if (info->no_rec < info->max_rec)
134 { /* there is room for this entry */
135 for (j = info->no_rec; j > i; --j)
136 info->sort_idx[j] = info->sort_idx[j-1];
137 idx = info->sort_idx[j] = info->no_rec;
141 return; /* score too low */
144 idx = info->sort_idx[0]; /* remove this entry */
147 for (j = 0; j < i; ++j) /* make room */
148 info->sort_idx[j] = info->sort_idx[j+1];
149 info->sort_idx[j] = idx; /* allocate sort entry */
151 memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
152 info->score_buf[idx] = score;
156 static struct rset_rel_info *qsort_info;
158 static int qcomp (const void *p1, const void *p2)
163 return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
164 qsort_info->key_buf + i2*qsort_info->key_size);
167 #define SCORE_SHOW 0.0 /* base score for showing up */
168 #define SCORE_COOC 0.3 /* component dependent on co-oc */
169 #define SCORE_DYN (1-(SCORE_SHOW+SCORE_COOC)) /* dynamic component of score */
171 static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
177 ISPT *isam_pt = NULL;
178 ISAMC_PP *isamc_pp = NULL;
181 logf (LOG_DEBUG, "relevance");
182 isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
183 isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
185 isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
187 isamc_pp = xmalloc (sizeof (*isamc_pp) * parms->no_isam_positions);
190 logf (LOG_FATAL, "No isamc or isam in rs_rel");
193 isam_tmp_buf = xmalloc (info->key_size);
194 max_tf = xmalloc (sizeof (*max_tf) * parms->no_terms);
195 tf = xmalloc (sizeof (*tf) * parms->no_terms);
197 for (i = 0; i<parms->no_terms; i++)
199 for (i = 0; i<parms->no_isam_positions; i++)
201 isam_buf[i] = xmalloc (info->key_size);
204 isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
205 max_tf [parms->term_no[i]] = is_numkeys (isam_pt[i]);
206 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
210 isamc_pp[i] = isc_pp_open (parms->isc, parms->isam_positions[i]);
211 max_tf [parms->term_no[i]] = isc_pp_num (isamc_pp[i]);
212 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
214 logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
221 for (i = 0; i<parms->no_isam_positions; i++)
224 (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 1))
229 pos = (*parms->get_pos)(isam_buf[i]);
230 logf (LOG_LOG, "pos=%d", pos);
232 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
234 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
241 int co_oc, last_term; /* Number of co-occurrences */
244 /* find min with lowest sysno */
245 for (i = 0; i<parms->no_isam_positions; i++)
248 (min < 0 || (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 2))
253 else if (!r && last_term != parms->term_no[i]) /* new occurrence */
255 last_term = parms->term_no[i];
260 memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
261 /* calculate for all with those sysno */
262 for (i = 0; i < parms->no_terms; i++)
264 for (i = 0; i<parms->no_isam_positions; i++)
269 r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
272 if (r <= 1 && r >= -1)
276 tf[parms->term_no[i]]++;
278 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
280 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
281 } while (isam_r[i] &&
282 (*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
285 /* calculate relevance value */
287 for (i = 0; i<parms->no_terms; i++)
289 score += SCORE_SHOW + SCORE_COOC*co_oc/parms->no_terms +
290 SCORE_DYN*tf[i]/max_tf[i];
291 /* if value is in the top score, then save it - don't emit yet */
292 add_rec (info, score/parms->no_terms, isam_tmp_buf);
295 for (i = 0; i<info->no_rec; i++)
296 info->sysno_idx[i] = i;
298 qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
299 for (i = 0; i<parms->no_isam_positions; i++)
302 is_pt_free (isam_pt[i]);
304 isc_pp_close (isamc_pp[i]);
308 xfree (isam_tmp_buf);
316 static void *r_create (const struct rset_control *sel, void *parms,
319 rset_relevance_parms *r_parms = parms;
320 struct rset_rel_info *info;
322 *flags |= RSET_FLAG_RANKED;
323 info = xmalloc (sizeof(struct rset_rel_info));
324 info->key_size = r_parms->key_size;
325 assert (info->key_size > 1);
326 info->max_rec = r_parms->max_rec;
327 assert (info->max_rec > 1);
328 info->cmp = r_parms->cmp;
329 info->get_pos = r_parms->get_pos;
331 info->key_buf = xmalloc (info->key_size * info->max_rec);
332 info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
333 info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
334 info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
336 info->rfd_list = NULL;
338 relevance (info, r_parms);
342 static RSFD r_open (RSET ct, int flag)
344 struct rset_rel_rfd *rfd;
345 struct rset_rel_info *info = ct->buf;
347 if (flag & RSETF_WRITE)
349 logf (LOG_FATAL, "relevance set type is read-only");
352 rfd = xmalloc (sizeof(*rfd));
354 rfd->next = info->rfd_list;
356 info->rfd_list = rfd;
361 static void r_close (RSFD rfd)
363 struct rset_rel_info *info = ((struct rset_rel_rfd*)rfd)->info;
364 struct rset_rel_rfd **rfdp;
366 for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
369 *rfdp = (*rfdp)->next;
373 logf (LOG_FATAL, "r_close but no rfd match!");
377 static void r_delete (RSET ct)
379 struct rset_rel_info *info = ct->buf;
381 assert (info->rfd_list == NULL);
382 xfree (info->key_buf);
383 xfree (info->score_buf);
384 xfree (info->sort_idx);
385 xfree (info->sysno_idx);
389 static void r_rewind (RSFD rfd)
391 struct rset_rel_rfd *p = rfd;
392 struct rset_rel_info *info = p->info;
394 if (p->flag & RSETF_SORT_RANK)
395 p->position = info->no_rec;
400 static int r_count (RSET ct)
402 struct rset_rel_info *info = ct->buf;
407 static int r_read (RSFD rfd, void *buf)
409 struct rset_rel_rfd *p = rfd;
410 struct rset_rel_info *info = p->info;
412 if (p->flag & RSETF_SORT_RANK)
414 if (p->position <= 0)
417 p->last_read_pos = info->sort_idx[p->position];
421 if (p->position == info->no_rec)
423 p->last_read_pos = info->sysno_idx[p->position];
427 info->key_buf + info->key_size * p->last_read_pos,
432 static int r_score (RSFD rfd, int *score)
434 struct rset_rel_rfd *p = rfd;
435 struct rset_rel_info *info = p->info;
437 *score = (int) (1000*info->score_buf[p->last_read_pos]);
441 static int r_write (RSFD rfd, const void *buf)
443 logf (LOG_FATAL, "relevance set type is read-only");