Mercurial > hg > camir-aes2014

annotate core/magnatagatune/AnnotDB.m @ 0:e9a9cd732c1e tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
first hg version after svn
author wolffd
date Tue, 10 Feb 2015 15:05:51 +0000
parents
children
rev   line source
wolffd@0 1 % The thesaurus class is a basic component of all
wolffd@0 2 % genre and tag information managing the whole Vocabulary.
wolffd@0 3
wolffd@0 4 classdef AnnotDB < handle
wolffd@0 5
wolffd@0 6 % public properties
wolffd@0 7 properties (SetAccess = private)
wolffd@0 8 lexicon = {};
wolffd@0 9
wolffd@0 10 end
wolffd@0 11
wolffd@0 12 properties(Hidden, Access = private)
wolffd@0 13
wolffd@0 14 annotsdb; % a numowners x numannots sparse binary / prob matrix
wolffd@0 15 annots_oid; % ownerid to pos in annots conversion
wolffd@0 16
wolffd@0 17 binary = 0; % indicator whether the db contains binary or scored annots
wolffd@0 18 end
wolffd@0 19
wolffd@0 20 methods
wolffd@0 21
wolffd@0 22 % ---
wolffd@0 23 % simple constructor
wolffd@0 24 % ---
wolffd@0 25 function db = AnnotDB(lexicon, annots, ids)
wolffd@0 26 % db = AnnotDB(lexicon, annots, annotation_ids)
wolffd@0 27 % lexicon: the list of all individual annotation elements or
wolffd@0 28 % clips_by_annot or
wolffd@0 29 % lexids_by_clip
wolffd@0 30
wolffd@0 31 % annots: either a clips x numel(lexicon) binary matrix or
wolffd@0 32 % ...
wolffd@0 33 %
wolffd@0 34 % annotation_ids: clip ids for the binary case,
wolffd@0 35
wolffd@0 36 if nargin >= 1
wolffd@0 37
wolffd@0 38 % ---
wolffd@0 39 % NOTE: two ways of supplying the annots are allowed:
wolffd@0 40 % 1. clip ids for each lexical element
wolffd@0 41 % 2. binary matrix
wolffd@0 42 % ---
wolffd@0 43 if ischar(lexicon)
wolffd@0 44
wolffd@0 45 if strcmp(lexicon, 'clips_by_annot')
wolffd@0 46
wolffd@0 47 % ---
wolffd@0 48 % preset the lexicon and hash ids
wolffd@0 49 % ---
wolffd@0 50 db.lexicon = unique(lower(annots));
wolffd@0 51
wolffd@0 52 if iscell(ids)
wolffd@0 53 db.annots_oid = unique([ids{:}]);
wolffd@0 54 else
wolffd@0 55 db.annots_oid = unique(ids);
wolffd@0 56 end
wolffd@0 57
wolffd@0 58 db.annotsdb = sparse(numel(db.annots_oid),...
wolffd@0 59 numel( db.lexicon));
wolffd@0 60
wolffd@0 61 % for all annotations
wolffd@0 62 for i = 1:numel(annots)
wolffd@0 63
wolffd@0 64 % for all ids in set
wolffd@0 65 % is this a cell or just a single index<
wolffd@0 66 if iscell(ids)
wolffd@0 67 for j = 1:numel(ids{i})
wolffd@0 68
wolffd@0 69 db.add_pair(ids{i}(j), annots{i});
wolffd@0 70 end
wolffd@0 71 else
wolffd@0 72 % single ndex case
wolffd@0 73 db.add_pair(ids(i), annots{i});
wolffd@0 74 end
wolffd@0 75 end
wolffd@0 76
wolffd@0 77 elseif strcmp(lexicon, 'annots_by_clip')
wolffd@0 78
wolffd@0 79 end
wolffd@0 80 % this is the binary case
wolffd@0 81 else
wolffd@0 82
wolffd@0 83 db.lexicon = lexicon;
wolffd@0 84 db.annotsdb = sparse(0,0);
wolffd@0 85 if nargin >= 2
wolffd@0 86
wolffd@0 87 db.annotsdb = sparse(annots);
wolffd@0 88 db.annots_oid = ids;
wolffd@0 89 else
wolffd@0 90 db.annotsdb = sparse(0, numel(db.lexicon));
wolffd@0 91 end
wolffd@0 92 end
wolffd@0 93 end
wolffd@0 94 end
wolffd@0 95
wolffd@0 96 % ---
wolffd@0 97 % retrieve annot-substructure for given clip ids,
wolffd@0 98 % collecting std = [or = all] ,[and = common]
wolffd@0 99 % annots for these
wolffd@0 100 % ---
wolffd@0 101 function new_db = subset(db, ownerids, mode)
wolffd@0 102 % new_db = subset(db, ownerids, {'and', ['or']})
wolffd@0 103
wolffd@0 104 if nargin < 3
wolffd@0 105 mode = 'or';
wolffd@0 106 end
wolffd@0 107
wolffd@0 108 % ---
wolffd@0 109 % create new DB
wolffd@0 110 % we make sure the tag id index keeps
wolffd@0 111 % the same for subsets by copying the whole
wolffd@0 112 % lexicon
wolffd@0 113 % ---
wolffd@0 114 new_db = AnnotDB(db.lexicon);
wolffd@0 115
wolffd@0 116 switch lower(mode)
wolffd@0 117 case 'and'
wolffd@0 118
wolffd@0 119 % ---
wolffd@0 120 % TODO: implement this and
wolffd@0 121 % improve speed below
wolffd@0 122 % ---
wolffd@0 123 case 'or'
wolffd@0 124
wolffd@0 125 % successively fill with given annots
wolffd@0 126 for i = 1:numel(ownerids)
wolffd@0 127
wolffd@0 128 % ---
wolffd@0 129 % we retrieve annots for each clip
wolffd@0 130 % and add them to the new database
wolffd@0 131 % ---
wolffd@0 132 [annot, score] = annots(db, ownerids(i));
wolffd@0 133 for j = 1:numel(annot)
wolffd@0 134
wolffd@0 135 new_db.add_pair(ownerids(i), annot{j}, score(j));
wolffd@0 136 end
wolffd@0 137 end
wolffd@0 138 otherwise
wolffd@0 139 error 'illegal owner id combination mode. possibly forgot brackets';
wolffd@0 140 end
wolffd@0 141 end
wolffd@0 142
wolffd@0 143 % retrieve annot-substructure for complement
wolffd@0 144 % of given clip ids
wolffd@0 145 function [new_db] = exclude(db, ownerids)
wolffd@0 146
wolffd@0 147 % get complement of clip ids
wolffd@0 148 ownerids = setdiff(db.annots_oid, ownerids);
wolffd@0 149
wolffd@0 150 new_db = subset(db, ownerids);
wolffd@0 151 end
wolffd@0 152
wolffd@0 153 % ---
wolffd@0 154 % retrieve clip by annot.
wolffd@0 155 % if multiple annots are given, the clips
wolffd@0 156 % containing all of them (logical and) are
wolffd@0 157 % returned
wolffd@0 158 % ---
wolffd@0 159 function oids = owner(db, annotstr, mode)
wolffd@0 160
wolffd@0 161 if nargin < 3
wolffd@0 162 mode = 'and';
wolffd@0 163 end
wolffd@0 164
wolffd@0 165 if ~iscell(annotstr)
wolffd@0 166 annotstr = {annotstr};
wolffd@0 167 end
wolffd@0 168
wolffd@0 169 annotid = [];
wolffd@0 170 for i = 1:numel(annotstr)
wolffd@0 171
wolffd@0 172 annotid = [annotid strcellfind(db.lexicon, annotstr{i})];
wolffd@0 173 end
wolffd@0 174
wolffd@0 175 oids = owner_for_annotid(db, annotid, mode);
wolffd@0 176 end
wolffd@0 177
wolffd@0 178 % retrieve owner ids by clip
wolffd@0 179 function ownerids = owner_for_annotid(db, annotid, mode)
wolffd@0 180 % ownerids = ownerids_for_annotid(db, annotid, {['and'], 'or'})
wolffd@0 181
wolffd@0 182 if isempty(annotid)
wolffd@0 183 ownerids = [];
wolffd@0 184 return
wolffd@0 185 end
wolffd@0 186 if nargin < 3
wolffd@0 187 mode = 'and';
wolffd@0 188 end
wolffd@0 189
wolffd@0 190 switch lower(mode)
wolffd@0 191 case 'or'
wolffd@0 192 % search for all appearing owners
wolffd@0 193 candidates = sum(db.annotsdb(:, annotid), 2) > 0;
wolffd@0 194
wolffd@0 195 case 'and'
wolffd@0 196 % search for the common owners
wolffd@0 197 candidates = sum(db.annotsdb(:, annotid), 2) == ...
wolffd@0 198 numel(annotid);
wolffd@0 199 otherwise
wolffd@0 200 error 'illegal tag combination mode';
wolffd@0 201 end
wolffd@0 202
wolffd@0 203
wolffd@0 204 % get positions in database
wolffd@0 205 pos = find(candidates);
wolffd@0 206
wolffd@0 207 % return owner ids
wolffd@0 208 ownerids = db.annots_oid(pos);
wolffd@0 209 end
wolffd@0 210
wolffd@0 211 % retrieve annotid by clip
wolffd@0 212 function [aid, score] = annotids_for_owner(db, ownerid, mode)
wolffd@0 213
wolffd@0 214 % single query case
wolffd@0 215 if numel(ownerid) == 1
wolffd@0 216
wolffd@0 217 pos = owner_pos(db, ownerid);
wolffd@0 218
wolffd@0 219 % get positions in database
wolffd@0 220 aid = find(db.annotsdb(pos, :) > 0);
wolffd@0 221
wolffd@0 222 score = db.annotsdb(pos, aid);
wolffd@0 223
wolffd@0 224
wolffd@0 225 % sort ids for output
wolffd@0 226 if ~db.binary
wolffd@0 227
wolffd@0 228 [score, idx] = sort(score, 'descend');
wolffd@0 229 aid = aid(idx);
wolffd@0 230 end
wolffd@0 231 else
wolffd@0 232 if nargin < 3
wolffd@0 233 mode = 'or';
wolffd@0 234 end
wolffd@0 235
wolffd@0 236 % ---
wolffd@0 237 % the query contained multiple ids
wolffd@0 238 %
wolffd@0 239 % we dont return the single results but
wolffd@0 240 % the statistics for this subset of clips
wolffd@0 241 % ---
wolffd@0 242 new_db = db.subset(ownerid, mode);
wolffd@0 243 [null, score, aid] = new_db.stats_count();
wolffd@0 244
wolffd@0 245 % cut off at score > 0 to abandon unused tags
wolffd@0 246 u = find(score > 0,1,'last');
wolffd@0 247 score = score(1:u);
wolffd@0 248 aid = aid(1:u);
wolffd@0 249 end
wolffd@0 250 end
wolffd@0 251
wolffd@0 252 % retrieve annotation by clip
wolffd@0 253 function [out, score, aid] = annots(db, ownerid)
wolffd@0 254
wolffd@0 255 [aid, score] = db.annotids_for_owner( ownerid);
wolffd@0 256
wolffd@0 257 out = db.get_annot_name(aid);
wolffd@0 258 end
wolffd@0 259
wolffd@0 260
wolffd@0 261 % retrieve annot name given a annot id
wolffd@0 262 function out = get_annot_name(db, annotid)
wolffd@0 263
wolffd@0 264 out = {};
wolffd@0 265 for i = 1:numel(annotid)
wolffd@0 266
wolffd@0 267 out{i} = db.lexicon{annotid(i)};
wolffd@0 268 end
wolffd@0 269 end
wolffd@0 270
wolffd@0 271 % return annotation id for annotation string
wolffd@0 272 function aid = get_annot_id(db, annotstr)
wolffd@0 273
wolffd@0 274 if ~iscell(annotstr)
wolffd@0 275
wolffd@0 276 % expensive search within annot list
wolffd@0 277 aid = strcellfind(db.lexicon, annotstr);
wolffd@0 278 else
wolffd@0 279
wolffd@0 280 % search seperately for each annot
wolffd@0 281 for i = 1:numel(annotstr)
wolffd@0 282 aid(i) = strcellfind(db.lexicon, annotstr{i});
wolffd@0 283 end
wolffd@0 284 end
wolffd@0 285 end
wolffd@0 286
wolffd@0 287 % ---
wolffd@0 288 % return statistics on saved annotations.
wolffd@0 289 % = returns the sum of the scores and
wolffd@0 290 % sortec lexicon
wolffd@0 291 % ---
wolffd@0 292 function [labels, score, annotids] = stats(db)
wolffd@0 293
wolffd@0 294 % out = zeros(1, size(db.annotsdb,2));
wolffd@0 295 score = full(sum(db.annotsdb, 1));
wolffd@0 296 [score, annotids] = sort(score,'descend');
wolffd@0 297
wolffd@0 298 % prepare labels
wolffd@0 299 labels = db.lexicon(annotids);
wolffd@0 300 end
wolffd@0 301
wolffd@0 302 % ---
wolffd@0 303 % return statistics on saved annotations.
wolffd@0 304 % = returns the number of annotations and
wolffd@0 305 % sortec lexicon
wolffd@0 306 % ---
wolffd@0 307 function [labels, score, annotids] = stats_count(db)
wolffd@0 308
wolffd@0 309 % out = zeros(1, size(db.annotsdb,2));
wolffd@0 310 score = full(sum(db.annotsdb > 0, 1));
wolffd@0 311 [score, annotids] = sort(score,'descend');
wolffd@0 312
wolffd@0 313 % prepare labels
wolffd@0 314 labels = db.lexicon(annotids);
wolffd@0 315 end
wolffd@0 316
wolffd@0 317 % this is a stub for a tag cloud-like output
wolffd@0 318 function [out] = annots_cloud(db, ownerid)
wolffd@0 319
wolffd@0 320 % ---
wolffd@0 321 % TODO: actually output tag-cloud
wolffd@0 322 % this output is aimed at input into a web interface
wolffd@0 323 % we successfully used http://www.wordle.net/
wolffd@0 324 % ---
wolffd@0 325
wolffd@0 326 if nargin > 1
wolffd@0 327 db2 = db.subset(ownerid);
wolffd@0 328 else
wolffd@0 329 db2 = db;
wolffd@0 330 end
wolffd@0 331
wolffd@0 332 [labels, score, annotids] = stats(db2);
wolffd@0 333
wolffd@0 334 % ---
wolffd@0 335 % Note: for performance issues we compress this data
wolffd@0 336 % to a maximum value of 1001
wolffd@0 337 % ---
wolffd@0 338 score = ceil((score./max(score))*100);
wolffd@0 339
wolffd@0 340 out = '';
wolffd@0 341 for i = 1:numel(annotids)
wolffd@0 342
wolffd@0 343 % repeat the tag according to score
wolffd@0 344 annot = strrep(labels{i},' ','-');
wolffd@0 345 for j = 1:score(i)
wolffd@0 346 out = sprintf('%s; %s',annot, out);
wolffd@0 347 end
wolffd@0 348 end
wolffd@0 349 end
wolffd@0 350
wolffd@0 351
wolffd@0 352 function out = size(db)
wolffd@0 353 % returns the size of this db
wolffd@0 354
wolffd@0 355 out = numel(db.lexicon);
wolffd@0 356 end
wolffd@0 357
wolffd@0 358 function add_pair(db, ownerid, annot, score)
wolffd@0 359 % add_pair(db, owner, annot) adds an annot and owner and can
wolffd@0 360 % increase the lexicon size
wolffd@0 361
wolffd@0 362 if nargin < 4
wolffd@0 363 score = 1;
wolffd@0 364 end
wolffd@0 365
wolffd@0 366 aid = strcellfind(db.lexicon, annot);
wolffd@0 367
wolffd@0 368 % create new position for annotation if neccesary
wolffd@0 369 if isempty(aid)
wolffd@0 370
wolffd@0 371 aid = numel(db.lexicon) + 1;
wolffd@0 372
wolffd@0 373 % add to lexicon
wolffd@0 374 db.lexicon = {db.lexicon{:}, annot};
wolffd@0 375
wolffd@0 376 % enhance annotation matrix
wolffd@0 377 db.annotsdb = [db.annotsdb, ...
wolffd@0 378 sparse(size(db.annotsdb,1), 1)];
wolffd@0 379 end
wolffd@0 380
wolffd@0 381
wolffd@0 382 % create new position for clip if neccesary
wolffd@0 383 pos = owner_pos(db, ownerid);
wolffd@0 384 if isempty(pos)
wolffd@0 385
wolffd@0 386 pos = numel(db.annots_oid) +1;
wolffd@0 387
wolffd@0 388 % add to oid
wolffd@0 389 db.annots_oid = [db.annots_oid, ownerid];
wolffd@0 390
wolffd@0 391 % enhance annotation matrix
wolffd@0 392 db.annotsdb = [db.annotsdb; ...
wolffd@0 393 sparse(1, size(db.annotsdb, 2))];
wolffd@0 394 end
wolffd@0 395
wolffd@0 396 % save data to database
wolffd@0 397 db.annotsdb(pos, aid) = score;
wolffd@0 398 end
wolffd@0 399
wolffd@0 400 end
wolffd@0 401
wolffd@0 402
wolffd@0 403 methods(Hidden)
wolffd@0 404
wolffd@0 405 function pos = owner_pos(db, ownerid)
wolffd@0 406
wolffd@0 407 % returns database position for owner id
wolffd@0 408 pos = find(db.annots_oid == ownerid);
wolffd@0 409 end
wolffd@0 410 end
wolffd@0 411
wolffd@0 412 end