2 * Copyright (C) 1994-1995, Index Data I/S
4 * Sebastian Hammer, Adam Dickmeiss
7 * Revision 1.7 1995-10-12 12:41:57 adam
8 * Private info (buf) moved from struct rset_control to struct rset.
9 * Bug fixes in relevance.
11 * Revision 1.6 1995/10/10 14:00:04 adam
12 * Function rset_open changed its wflag parameter to general flags.
14 * Revision 1.5 1995/10/06 14:38:06 adam
15 * New result set method: r_score.
16 * Local no (sysno) and score is transferred to retrieveCtrl.
18 * Revision 1.4 1995/09/14 07:48:56 adam
19 * Other score calculation.
21 * Revision 1.3 1995/09/11 15:23:40 adam
22 * More work on relevance search.
24 * Revision 1.2 1995/09/11 13:09:41 adam
25 * More work on relevance feedback.
27 * Revision 1.1 1995/09/08 14:52:42 adam
28 * Work on relevance feedback.
41 static void *r_create(const struct rset_control *sel, void *parms);
42 static RSFD r_open (RSET ct, int flag);
43 static void r_close (RSFD rfd);
44 static void r_delete (RSET ct);
45 static void r_rewind (RSFD rfd);
46 static int r_count (RSET ct);
47 static int r_read (RSFD rfd, void *buf);
48 static int r_write (RSFD rfd, const void *buf);
49 static int r_score (RSFD rfd, int *score);
51 static const rset_control control =
65 const rset_control *rset_kind_relevance = &control;
67 struct rset_rel_info {
71 int (*cmp)(const void *p1, const void *p2);
72 char *key_buf; /* key buffer */
73 float *score_buf; /* score buffer */
74 int *sort_idx; /* score sorted index */
75 int *sysno_idx; /* sysno sorted index (ring buffer) */
76 struct rset_rel_rfd *rfd_list;
83 struct rset_rel_rfd *next;
84 struct rset_rel_info *info;
87 static void add_rec (struct rset_rel_info *info, double score, void *key)
91 for (i = 0; i<info->no_rec; i++)
93 idx = info->sort_idx[i];
94 if (score <= info->score_buf[idx])
97 if (info->no_rec < info->max_rec)
98 { /* there is room for this entry */
99 for (j = info->no_rec; j > i; --j)
100 info->sort_idx[j] = info->sort_idx[j-1];
101 idx = info->sort_idx[j] = info->no_rec;
105 return; /* score too low */
108 idx = info->sort_idx[0]; /* remove this entry */
111 for (j = 0; j < i; ++j) /* make room */
112 info->sort_idx[j] = info->sort_idx[j+1];
113 info->sort_idx[j] = idx; /* allocate sort entry */
115 memcpy (info->key_buf + idx*info->key_size, key, info->key_size);
116 info->score_buf[idx] = score;
120 static struct rset_rel_info *qsort_info;
122 static int qcomp (const void *p1, const void *p2)
127 return qsort_info->cmp (qsort_info->key_buf + i1*qsort_info->key_size,
128 qsort_info->key_buf + i2*qsort_info->key_size);
131 static void relevance (struct rset_rel_info *info, rset_relevance_parms *parms)
141 logf (LOG_DEBUG, "relevance");
142 isam_buf = xmalloc (parms->no_isam_positions * sizeof(*isam_buf));
143 isam_r = xmalloc (sizeof (*isam_r) * parms->no_isam_positions);
144 isam_pt = xmalloc (sizeof (*isam_pt) * parms->no_isam_positions);
145 isam_tmp_buf = xmalloc (info->key_size);
146 max_tf = xmalloc (sizeof (*max_tf) * parms->no_isam_positions);
147 wgt = xmalloc (sizeof (*wgt) * parms->no_isam_positions);
149 for (i = 0; i<parms->no_isam_positions; i++)
151 isam_buf[i] = xmalloc (info->key_size);
152 isam_pt[i] = is_position (parms->is, parms->isam_positions[i]);
153 max_tf [i] = is_numkeys (isam_pt[i]);
154 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
155 logf (LOG_DEBUG, "max tf %d = %d", i, max_tf[i]);
162 /* find min with lowest sysno */
163 for (i = 0; i<parms->no_isam_positions; i++)
165 (min < 0 || (*parms->cmp)(isam_buf[i], isam_buf[min]) < 1))
169 memcpy (isam_tmp_buf, isam_buf[min], info->key_size);
170 /* calculate for all with those sysno */
171 for (i = 0; i<parms->no_isam_positions; i++)
176 r = (*parms->cmp)(isam_buf[i], isam_tmp_buf);
187 isam_r[i] = is_readkey (isam_pt[i], isam_buf[i]);
188 } while (isam_r[i] &&
189 (*parms->cmp)(isam_buf[i], isam_tmp_buf) <= 1);
190 wgt[i] = 0.1+tf*0.9/max_tf[i];
193 /* calculate relevance value */
195 for (i = 0; i<parms->no_isam_positions; i++)
197 /* if value is in the top score, then save it - don't emit yet */
198 add_rec (info, score, isam_tmp_buf);
200 for (i = 0; i<info->no_rec; i++)
201 info->sysno_idx[i] = i;
203 qsort (info->sysno_idx, info->no_rec, sizeof(*info->sysno_idx), qcomp);
204 for (i = 0; i<parms->no_isam_positions; i++)
206 is_pt_free (isam_pt[i]);
210 xfree (isam_tmp_buf);
217 static void *r_create (const struct rset_control *sel, void *parms)
219 rset_relevance_parms *r_parms = parms;
220 struct rset_rel_info *info;
222 info = xmalloc (sizeof(struct rset_rel_info));
223 info->key_size = r_parms->key_size;
224 assert (info->key_size > 1);
225 info->max_rec = r_parms->max_rec;
226 assert (info->max_rec > 1);
227 info->cmp = r_parms->cmp;
229 info->key_buf = xmalloc (info->key_size * info->max_rec);
230 info->score_buf = xmalloc (sizeof(*info->score_buf) * info->max_rec);
231 info->sort_idx = xmalloc (sizeof(*info->sort_idx) * info->max_rec);
232 info->sysno_idx = xmalloc (sizeof(*info->sysno_idx) * info->max_rec);
234 info->rfd_list = NULL;
236 relevance (info, r_parms);
240 static RSFD r_open (RSET ct, int flag)
242 struct rset_rel_rfd *rfd;
243 struct rset_rel_info *info = ct->buf;
245 if (flag & RSETF_WRITE)
247 logf (LOG_FATAL, "relevance set type is read-only");
250 rfd = xmalloc (sizeof(*rfd));
252 rfd->next = info->rfd_list;
254 info->rfd_list = rfd;
259 static void r_close (RSFD rfd)
261 struct rset_rel_info *info = ((struct rset_rel_rfd*)rfd)->info;
262 struct rset_rel_rfd **rfdp;
264 for (rfdp = &info->rfd_list; *rfdp; rfdp = &(*rfdp)->next)
267 *rfdp = (*rfdp)->next;
271 logf (LOG_FATAL, "r_close but no rfd match!");
275 static void r_delete (RSET ct)
277 struct rset_rel_info *info = ct->buf;
279 assert (info->rfd_list == NULL);
280 xfree (info->key_buf);
281 xfree (info->score_buf);
282 xfree (info->sort_idx);
283 xfree (info->sysno_idx);
288 static void r_rewind (RSFD rfd)
290 struct rset_rel_rfd *p = rfd;
291 struct rset_rel_info *info = p->info;
293 if (p->flag & RSETF_SORT_RANK)
294 p->position = info->no_rec;
299 static int r_count (RSET ct)
301 struct rset_rel_info *info = ct->buf;
306 static int r_read (RSFD rfd, void *buf)
308 struct rset_rel_rfd *p = rfd;
309 struct rset_rel_info *info = p->info;
311 if (p->flag & RSETF_SORT_RANK)
313 if (p->position <= 0)
316 p->last_read_pos = info->sort_idx[p->position];
320 if (p->position == info->no_rec)
322 p->last_read_pos = info->sysno_idx[p->position];
326 info->key_buf + info->key_size * p->last_read_pos,
331 static int r_score (RSFD rfd, int *score)
333 struct rset_rel_rfd *p = rfd;
334 struct rset_rel_info *info = p->info;
336 *score = (int) (1000*info->score_buf[p->last_read_pos]);
340 static int r_write (RSFD rfd, const void *buf)
342 logf (LOG_FATAL, "relevance set type is read-only");