2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.15 1997-09-09 13:38:16 adam
8 * Partial port to WIN95/NT.
10 * Revision 1.14 1996/11/08 11:15:58 adam
11 * Compressed isam fully supported.
13 * Revision 1.13 1996/10/29 13:55:26 adam
14 * Include of zebrautl.h instead of alexutil.h.
16 * Revision 1.12 1996/10/08 13:00:40 adam
17 * Bug fix: result sets with ranked operands in boolean operations weren't
20 * Revision 1.11 1996/10/07 16:05:29 quinn
23 * Revision 1.9 1995/12/11 09:15:26 adam
24 * New set types: sand/sor/snot - ranked versions of and/or/not in
25 * ranked/semi-ranked result sets.
26 * Note: the snot not finished yet.
27 * New rset member: flag.
28 * Bug fix: r_delete in rsrel.c did free bad memory block.
30 * Revision 1.8 1995/12/05 11:25:45 adam
31 * Doesn't include math.h.
33 * Revision 1.7 1995/10/12 12:41:57 adam
34 * Private info (buf) moved from struct rset_control to struct rset.
35 * Bug fixes in relevance.
37 * Revision 1.6 1995/10/10 14:00:04 adam
38 * Function rset_open changed its wflag parameter to general flags.
40 * Revision 1.5 1995/10/06 14:38:06 adam
41 * New result set method: r_score.
42 * Local no (sysno) and score is transferred to retrieveCtrl.
44 * Revision 1.4 1995/09/14 07:48:56 adam
45 * Other score calculation.
47 * Revision 1.3 1995/09/11 15:23:40 adam
48 * More work on relevance search.
50 * Revision 1.2 1995/09/11 13:09:41 adam
51 * More work on relevance feedback.
53 * Revision 1.1 1995/09/08 14:52:42 adam
54 * Work on relevance feedback.
68 static void *r_create(const struct rset_control *sel, void *parms,
70 static RSFD r_open (RSET ct, int flag);
71 static void r_close (RSFD rfd);
72 static void r_delete (RSET ct);
73 static void r_rewind (RSFD rfd);
74 static int r_count (RSET ct);
75 static int r_read (RSFD rfd, void *buf);
76 static int r_write (RSFD rfd, const void *buf);
77 static int r_score (RSFD rfd, int *score);
79 static const rset_control control =
93 const rset_control *rset_kind_relevance = &control;
95 struct rset_rel_info {
99 int (*cmp)(const void *p1, const void *p2);
100 char *key_buf; /* key buffer */
101 float *score_buf; /* score buffer */
102 int *sort_idx; /* score sorted index */
103 int *sysno_idx; /* sysno sorted index (ring buffer) */
104 struct rset_rel_rfd *rfd_list;
107 struct rset_rel_rfd {
111 struct rset_rel_rfd *next;
112 struct rset_rel_info *info;
115 static void add_rec (struct rset_rel_info *info, double score, void *key)
119 for (i = 0; i<info->no_rec; i++)
121 idx = info->sort_idx[i];
122 if (score <= info->score_buf[idx])
125 if (info->no_rec < info->max_rec)
126 { /* there is room for this entry */
127 for (j = info->no_rec; j > i; --j)
128 info->sort_idx[j] = info->sort_idx[j-1];
129 idx = info->sort_idx[j] = info->no_rec;
133 return; /* score too low */
136 idx = info->sort_idx[0]; /* remove this entry */
139 for (j = 0; j < i; ++j) /* make room */
140 info->sort_idx[j] = info->sort_idx[j+1];
141 info->sort_idx[j] = idx; /* allocate sort entry */
143 memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
144 info->score_buf[idx] = score;
148 static struct rset_rel_info *qsort_info;
150 static int qcomp (const void *p1, const void *p2)
155 return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
156 qsort_info->key_buf + i2*qsort_info->key_size);
159 #define SCORE_SHOW 0.0 /* base score for showing up */
160 #define SCORE_COOC 0.3 /* component dependent on co-oc */
161 #define SCORE_DYN (1-(SCORE_SHOW+SCORE_COOC)) /* dynamic component of score */
163 static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
169 ISPT *isam_pt = NULL;
170 ISAMC_PP *isamc_pp = NULL;
173 logf (LOG_DEBUG, "relevance");
174 isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
175 isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
177 isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
179 isamc_pp = xmalloc (sizeof (*isamc_pp) * parms->no_isam_positions);
182 logf (LOG_FATAL, "No isamc or isam in rs_rel");
185 isam_tmp_buf = xmalloc (info->key_size);
186 max_tf = xmalloc (sizeof (*max_tf) * parms->no_terms);
187 tf = xmalloc (sizeof (*tf) * parms->no_terms);
189 for (i = 0; i<parms->no_terms; i++)
191 for (i = 0; i<parms->no_isam_positions; i++)
193 isam_buf[i] = xmalloc (info->key_size);
196 isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
197 max_tf [parms->term_no[i]] = is_numkeys (isam_pt[i]);
198 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
202 isamc_pp[i] = isc_pp_open (parms->isc, parms->isam_positions[i]);
203 max_tf [parms->term_no[i]] = isc_pp_num (isamc_pp[i]);
204 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
206 logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
212 int co_oc, last_term; /* Number of co-occurrences */
215 /* find min with lowest sysno */
216 for (i = 0; i<parms->no_isam_positions; i++)
219 (min < 0 || (r = (*parms->cmp)(isam_buf[i], isam_buf[min])) < 2))
224 else if (!r && last_term != parms->term_no[i]) /* new occurrence */
226 last_term = parms->term_no[i];
231 memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
232 /* calculate for all with those sysno */
233 for (i = 0; i < parms->no_terms; i++)
235 for (i = 0; i<parms->no_isam_positions; i++)
240 r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
243 if (r <= 1 && r >= -1)
247 tf[parms->term_no[i]]++;
249 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
251 isam_r[i] = isc_pp_read (isamc_pp[i], isam_buf[i]);
252 } while (isam_r[i] &&
253 (*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
256 /* calculate relevance value */
258 for (i = 0; i<parms->no_terms; i++)
260 score += SCORE_SHOW + SCORE_COOC*co_oc/parms->no_terms +
261 SCORE_DYN*tf[i]/max_tf[i];
262 /* if value is in the top score, then save it - don't emit yet */
263 add_rec (info, score/parms->no_terms, isam_tmp_buf);
265 for (i = 0; i<info->no_rec; i++)
266 info->sysno_idx[i] = i;
268 qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
269 for (i = 0; i<parms->no_isam_positions; i++)
272 is_pt_free (isam_pt[i]);
274 isc_pp_close (isamc_pp[i]);
278 xfree (isam_tmp_buf);
286 static void *r_create (const struct rset_control *sel, void *parms,
289 rset_relevance_parms *r_parms = parms;
290 struct rset_rel_info *info;
292 *flags |= RSET_FLAG_RANKED;
293 info = xmalloc (sizeof(struct rset_rel_info));
294 info->key_size = r_parms->key_size;
295 assert (info->key_size > 1);
296 info->max_rec = r_parms->max_rec;
297 assert (info->max_rec > 1);
298 info->cmp = r_parms->cmp;
300 info->key_buf = xmalloc (info->key_size * info->max_rec);
301 info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
302 info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
303 info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
305 info->rfd_list = NULL;
307 relevance (info, r_parms);
311 static RSFD r_open (RSET ct, int flag)
313 struct rset_rel_rfd *rfd;
314 struct rset_rel_info *info = ct->buf;
316 if (flag & RSETF_WRITE)
318 logf (LOG_FATAL, "relevance set type is read-only");
321 rfd = xmalloc (sizeof(*rfd));
323 rfd->next = info->rfd_list;
325 info->rfd_list = rfd;
330 static void r_close (RSFD rfd)
332 struct rset_rel_info *info = ((struct rset_rel_rfd*)rfd)->info;
333 struct rset_rel_rfd **rfdp;
335 for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
338 *rfdp = (*rfdp)->next;
342 logf (LOG_FATAL, "r_close but no rfd match!");
346 static void r_delete (RSET ct)
348 struct rset_rel_info *info = ct->buf;
350 assert (info->rfd_list == NULL);
351 xfree (info->key_buf);
352 xfree (info->score_buf);
353 xfree (info->sort_idx);
354 xfree (info->sysno_idx);
358 static void r_rewind (RSFD rfd)
360 struct rset_rel_rfd *p = rfd;
361 struct rset_rel_info *info = p->info;
363 if (p->flag & RSETF_SORT_RANK)
364 p->position = info->no_rec;
369 static int r_count (RSET ct)
371 struct rset_rel_info *info = ct->buf;
376 static int r_read (RSFD rfd, void *buf)
378 struct rset_rel_rfd *p = rfd;
379 struct rset_rel_info *info = p->info;
381 if (p->flag & RSETF_SORT_RANK)
383 if (p->position <= 0)
386 p->last_read_pos = info->sort_idx[p->position];
390 if (p->position == info->no_rec)
392 p->last_read_pos = info->sysno_idx[p->position];
396 info->key_buf + info->key_size * p->last_read_pos,
401 static int r_score (RSFD rfd, int *score)
403 struct rset_rel_rfd *p = rfd;
404 struct rset_rel_info *info = p->info;
406 *score = (int) (1000*info->score_buf[p->last_read_pos]);
410 static int r_write (RSFD rfd, const void *buf)
412 logf (LOG_FATAL, "relevance set type is read-only");