[13] | 1 | package de.ugoe.cs.eventbench.models;
|
---|
[12] | 2 |
|
---|
[93] | 3 | import java.security.InvalidParameterException;
|
---|
| 4 | import java.util.ArrayList;
|
---|
[102] | 5 | import java.util.Collection;
|
---|
[252] | 6 | import java.util.HashSet;
|
---|
[93] | 7 | import java.util.LinkedHashSet;
|
---|
[12] | 8 | import java.util.LinkedList;
|
---|
| 9 | import java.util.List;
|
---|
| 10 | import java.util.Random;
|
---|
[80] | 11 | import java.util.Set;
|
---|
[12] | 12 |
|
---|
| 13 | import de.ugoe.cs.eventbench.data.Event;
|
---|
[23] | 14 | import de.ugoe.cs.eventbench.models.Trie.Edge;
|
---|
| 15 | import de.ugoe.cs.eventbench.models.Trie.TrieVertex;
|
---|
| 16 | import edu.uci.ics.jung.graph.Tree;
|
---|
[12] | 17 |
|
---|
[100] | 18 | /**
|
---|
| 19 | * <p>
|
---|
| 20 | * Implements a skeleton for stochastic processes that can calculate
|
---|
| 21 | * probabilities based on a trie. The skeleton provides all functionalities of
|
---|
| 22 | * {@link IStochasticProcess} except
|
---|
| 23 | * {@link IStochasticProcess#getProbability(List, Event)}.
|
---|
| 24 | * </p>
|
---|
| 25 | *
|
---|
| 26 | * @author Steffen Herbold
|
---|
| 27 | * @version 1.0
|
---|
| 28 | */
|
---|
[17] | 29 | public abstract class TrieBasedModel implements IStochasticProcess {
|
---|
[12] | 30 |
|
---|
[86] | 31 | /**
|
---|
[100] | 32 | * <p>
|
---|
[86] | 33 | * Id for object serialization.
|
---|
[100] | 34 | * </p>
|
---|
[86] | 35 | */
|
---|
| 36 | private static final long serialVersionUID = 1L;
|
---|
| 37 |
|
---|
[100] | 38 | /**
|
---|
| 39 | * <p>
|
---|
| 40 | * The order of the trie, i.e., the maximum length of subsequences stored in
|
---|
| 41 | * the trie.
|
---|
| 42 | * </p>
|
---|
| 43 | */
|
---|
[16] | 44 | protected int trieOrder;
|
---|
[12] | 45 |
|
---|
[100] | 46 | /**
|
---|
| 47 | * <p>
|
---|
| 48 | * Trie on which the probability calculations are based.
|
---|
| 49 | * </p>
|
---|
| 50 | */
|
---|
[182] | 51 | protected Trie<Event<?>> trie = null;
|
---|
[100] | 52 |
|
---|
| 53 | /**
|
---|
| 54 | * <p>
|
---|
| 55 | * Random number generator used by probabilistic sequence generation
|
---|
| 56 | * methods.
|
---|
| 57 | * </p>
|
---|
| 58 | */
|
---|
[12] | 59 | protected final Random r;
|
---|
| 60 |
|
---|
[100] | 61 | /**
|
---|
| 62 | * <p>
|
---|
| 63 | * Constructor. Creates a new TrieBasedModel that can be used for stochastic
|
---|
| 64 | * processes with a Markov order less than or equal to {@code markovOrder}.
|
---|
| 65 | * </p>
|
---|
| 66 | *
|
---|
| 67 | * @param markovOrder
|
---|
| 68 | * Markov order of the model
|
---|
| 69 | * @param r
|
---|
| 70 | * random number generator used by probabilistic methods of the
|
---|
| 71 | * class
|
---|
[342] | 72 | * @throws InvalidParameterException
|
---|
| 73 | * thrown if markovOrder is less than 0 or the random number
|
---|
| 74 | * generator r is null
|
---|
[100] | 75 | */
|
---|
[16] | 76 | public TrieBasedModel(int markovOrder, Random r) {
|
---|
[12] | 77 | super();
|
---|
[342] | 78 | if (markovOrder < 0) {
|
---|
| 79 | throw new InvalidParameterException(
|
---|
| 80 | "markov order must not be less than 0");
|
---|
| 81 | }
|
---|
| 82 | if (r == null) {
|
---|
| 83 | throw new InvalidParameterException(
|
---|
| 84 | "random number generator r must not be null");
|
---|
| 85 | }
|
---|
[100] | 86 | this.trieOrder = markovOrder + 1;
|
---|
[12] | 87 | this.r = r;
|
---|
| 88 | }
|
---|
| 89 |
|
---|
[100] | 90 | /**
|
---|
| 91 | * <p>
|
---|
| 92 | * Trains the model by generating a trie from which probabilities are
|
---|
[182] | 93 | * calculated. The trie is newly generated based solely on the passed
|
---|
| 94 | * sequences. If an existing model should only be updated, use
|
---|
| 95 | * {@link #update(Collection)} instead.
|
---|
[100] | 96 | * </p>
|
---|
| 97 | *
|
---|
| 98 | * @param sequences
|
---|
| 99 | * training data
|
---|
[342] | 100 | * @throws InvalidParameterException
|
---|
| 101 | * thrown is sequences is null
|
---|
[100] | 102 | */
|
---|
[325] | 103 | public void train(Collection<List<? extends Event<?>>> sequences) {
|
---|
[182] | 104 | trie = null;
|
---|
| 105 | update(sequences);
|
---|
| 106 | }
|
---|
[100] | 107 |
|
---|
[182] | 108 | /**
|
---|
| 109 | * <p>
|
---|
| 110 | * Trains the model by updating the trie from which the probabilities are
|
---|
| 111 | * calculated. This function updates an existing trie. In case no trie
|
---|
| 112 | * exists yet, a new trie is generated and the function behaves like
|
---|
| 113 | * {@link #train(Collection)}.
|
---|
| 114 | * </p>
|
---|
| 115 | *
|
---|
| 116 | * @param sequences
|
---|
| 117 | * training data
|
---|
[342] | 118 | * @throws InvalidParameterException
|
---|
| 119 | * thrown is sequences is null
|
---|
[182] | 120 | */
|
---|
[325] | 121 | public void update(Collection<List<? extends Event<?>>> sequences) {
|
---|
[252] | 122 | if (sequences == null) {
|
---|
[342] | 123 | throw new InvalidParameterException("sequences must not be null");
|
---|
[252] | 124 | }
|
---|
[182] | 125 | if (trie == null) {
|
---|
| 126 | trie = new Trie<Event<?>>();
|
---|
| 127 | }
|
---|
[325] | 128 | for (List<? extends Event<?>> sequence : sequences) {
|
---|
[100] | 129 | List<Event<?>> currentSequence = new LinkedList<Event<?>>(sequence); // defensive
|
---|
| 130 | // copy
|
---|
[12] | 131 | currentSequence.add(0, Event.STARTEVENT);
|
---|
| 132 | currentSequence.add(Event.ENDEVENT);
|
---|
[100] | 133 |
|
---|
[16] | 134 | trie.train(currentSequence, trieOrder);
|
---|
[12] | 135 | }
|
---|
| 136 | }
|
---|
| 137 |
|
---|
[100] | 138 | /*
|
---|
| 139 | * (non-Javadoc)
|
---|
| 140 | *
|
---|
[17] | 141 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#randomSequence()
|
---|
| 142 | */
|
---|
| 143 | @Override
|
---|
[12] | 144 | public List<? extends Event<?>> randomSequence() {
|
---|
[386] | 145 | return randomSequence(Integer.MAX_VALUE, true);
|
---|
| 146 | }
|
---|
| 147 |
|
---|
| 148 | /*
|
---|
| 149 | * (non-Javadoc)
|
---|
| 150 | *
|
---|
| 151 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#randomSequence()
|
---|
| 152 | */
|
---|
| 153 | @Override
|
---|
| 154 | public List<? extends Event<?>> randomSequence(int maxLength,
|
---|
| 155 | boolean validEnd) {
|
---|
[12] | 156 | List<Event<?>> sequence = new LinkedList<Event<?>>();
|
---|
[342] | 157 | if (trie != null) {
|
---|
[386] | 158 | boolean endFound = false;
|
---|
| 159 | while (!endFound) { // outer loop for length checking
|
---|
| 160 | sequence = new LinkedList<Event<?>>();
|
---|
| 161 | IncompleteMemory<Event<?>> context = new IncompleteMemory<Event<?>>(
|
---|
| 162 | trieOrder - 1);
|
---|
| 163 | context.add(Event.STARTEVENT);
|
---|
[342] | 164 |
|
---|
[386] | 165 | while (!endFound && sequence.size() < maxLength) {
|
---|
| 166 | double randVal = r.nextDouble();
|
---|
| 167 | double probSum = 0.0;
|
---|
| 168 | List<Event<?>> currentContext = context.getLast(trieOrder);
|
---|
| 169 | for (Event<?> symbol : trie.getKnownSymbols()) {
|
---|
| 170 | probSum += getProbability(currentContext, symbol);
|
---|
| 171 | if (probSum >= randVal) {
|
---|
| 172 | if (!(symbol == Event.STARTEVENT || symbol == Event.ENDEVENT)) {
|
---|
| 173 | // only add the symbol the sequence if it is not
|
---|
| 174 | // START or END
|
---|
| 175 | context.add(symbol);
|
---|
[391] | 176 | sequence.add(symbol);
|
---|
[386] | 177 | }
|
---|
| 178 | endFound = (symbol == Event.ENDEVENT)
|
---|
| 179 | || (!validEnd && sequence.size() == maxLength);
|
---|
| 180 | break;
|
---|
[252] | 181 | }
|
---|
[12] | 182 | }
|
---|
| 183 | }
|
---|
| 184 | }
|
---|
| 185 | }
|
---|
| 186 | return sequence;
|
---|
| 187 | }
|
---|
[100] | 188 |
|
---|
| 189 | /**
|
---|
| 190 | * <p>
|
---|
| 191 | * Returns a Dot representation of the internal trie.
|
---|
| 192 | * </p>
|
---|
| 193 | *
|
---|
| 194 | * @return dot representation of the internal trie
|
---|
| 195 | */
|
---|
[30] | 196 | public String getTrieDotRepresentation() {
|
---|
[252] | 197 | if (trie == null) {
|
---|
| 198 | return "";
|
---|
| 199 | } else {
|
---|
| 200 | return trie.getDotRepresentation();
|
---|
| 201 | }
|
---|
[30] | 202 | }
|
---|
[100] | 203 |
|
---|
| 204 | /**
|
---|
| 205 | * <p>
|
---|
| 206 | * Returns a {@link Tree} of the internal trie that can be used for
|
---|
| 207 | * visualization.
|
---|
| 208 | * </p>
|
---|
| 209 | *
|
---|
| 210 | * @return {@link Tree} depicting the internal trie
|
---|
| 211 | */
|
---|
[23] | 212 | public Tree<TrieVertex, Edge> getTrieGraph() {
|
---|
[252] | 213 | if (trie == null) {
|
---|
| 214 | return null;
|
---|
| 215 | } else {
|
---|
| 216 | return trie.getGraph();
|
---|
| 217 | }
|
---|
[23] | 218 | }
|
---|
[12] | 219 |
|
---|
[100] | 220 | /**
|
---|
| 221 | * <p>
|
---|
| 222 | * The string representation of the model is {@link Trie#toString()} of
|
---|
| 223 | * {@link #trie}.
|
---|
| 224 | * </p>
|
---|
| 225 | *
|
---|
| 226 | * @see java.lang.Object#toString()
|
---|
| 227 | */
|
---|
[12] | 228 | @Override
|
---|
| 229 | public String toString() {
|
---|
[252] | 230 | if (trie == null) {
|
---|
| 231 | return "";
|
---|
| 232 | } else {
|
---|
| 233 | return trie.toString();
|
---|
| 234 | }
|
---|
[12] | 235 | }
|
---|
[100] | 236 |
|
---|
| 237 | /*
|
---|
| 238 | * (non-Javadoc)
|
---|
| 239 | *
|
---|
| 240 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#getNumStates()
|
---|
| 241 | */
|
---|
| 242 | @Override
|
---|
[129] | 243 | public int getNumSymbols() {
|
---|
[252] | 244 | if (trie == null) {
|
---|
| 245 | return 0;
|
---|
| 246 | } else {
|
---|
| 247 | return trie.getNumSymbols();
|
---|
| 248 | }
|
---|
[66] | 249 | }
|
---|
[100] | 250 |
|
---|
| 251 | /*
|
---|
| 252 | * (non-Javadoc)
|
---|
| 253 | *
|
---|
| 254 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#getStateStrings()
|
---|
| 255 | */
|
---|
| 256 | @Override
|
---|
[129] | 257 | public String[] getSymbolStrings() {
|
---|
[252] | 258 | if (trie == null) {
|
---|
| 259 | return new String[0];
|
---|
| 260 | }
|
---|
[129] | 261 | String[] stateStrings = new String[getNumSymbols()];
|
---|
[100] | 262 | int i = 0;
|
---|
| 263 | for (Event<?> symbol : trie.getKnownSymbols()) {
|
---|
[386] | 264 | if (symbol.toString() == null) {
|
---|
| 265 | stateStrings[i] = "null";
|
---|
| 266 | } else {
|
---|
| 267 | stateStrings[i] = symbol.toString();
|
---|
| 268 | }
|
---|
[70] | 269 | i++;
|
---|
| 270 | }
|
---|
| 271 | return stateStrings;
|
---|
| 272 | }
|
---|
[100] | 273 |
|
---|
| 274 | /*
|
---|
| 275 | * (non-Javadoc)
|
---|
| 276 | *
|
---|
| 277 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#getEvents()
|
---|
| 278 | */
|
---|
| 279 | @Override
|
---|
[102] | 280 | public Collection<? extends Event<?>> getEvents() {
|
---|
[252] | 281 | if (trie == null) {
|
---|
| 282 | return new HashSet<Event<?>>();
|
---|
| 283 | } else {
|
---|
| 284 | return trie.getKnownSymbols();
|
---|
| 285 | }
|
---|
[80] | 286 | }
|
---|
[100] | 287 |
|
---|
| 288 | /*
|
---|
| 289 | * (non-Javadoc)
|
---|
| 290 | *
|
---|
| 291 | * @see
|
---|
| 292 | * de.ugoe.cs.eventbench.models.IStochasticProcess#generateSequences(int)
|
---|
| 293 | */
|
---|
| 294 | @Override
|
---|
[102] | 295 | public Collection<List<? extends Event<?>>> generateSequences(int length) {
|
---|
[94] | 296 | return generateSequences(length, false);
|
---|
[93] | 297 | }
|
---|
[100] | 298 |
|
---|
| 299 | /*
|
---|
| 300 | * (non-Javadoc)
|
---|
| 301 | *
|
---|
| 302 | * @see
|
---|
| 303 | * de.ugoe.cs.eventbench.models.IStochasticProcess#generateSequences(int,
|
---|
| 304 | * boolean)
|
---|
| 305 | */
|
---|
| 306 | @Override
|
---|
| 307 | public Set<List<? extends Event<?>>> generateSequences(int length,
|
---|
| 308 | boolean fromStart) {
|
---|
| 309 | Set<List<? extends Event<?>>> sequenceSet = new LinkedHashSet<List<? extends Event<?>>>();
|
---|
| 310 | if (length < 1) {
|
---|
| 311 | throw new InvalidParameterException(
|
---|
| 312 | "Length of generated subsequences must be at least 1.");
|
---|
[94] | 313 | }
|
---|
[100] | 314 | if (length == 1) {
|
---|
| 315 | if (fromStart) {
|
---|
[94] | 316 | List<Event<?>> subSeq = new LinkedList<Event<?>>();
|
---|
| 317 | subSeq.add(Event.STARTEVENT);
|
---|
[95] | 318 | sequenceSet.add(subSeq);
|
---|
[94] | 319 | } else {
|
---|
[100] | 320 | for (Event<?> event : getEvents()) {
|
---|
[94] | 321 | List<Event<?>> subSeq = new LinkedList<Event<?>>();
|
---|
| 322 | subSeq.add(event);
|
---|
| 323 | sequenceSet.add(subSeq);
|
---|
| 324 | }
|
---|
| 325 | }
|
---|
| 326 | return sequenceSet;
|
---|
| 327 | }
|
---|
[102] | 328 | Collection<? extends Event<?>> events = getEvents();
|
---|
| 329 | Collection<List<? extends Event<?>>> seqsShorter = generateSequences(
|
---|
[100] | 330 | length - 1, fromStart);
|
---|
| 331 | for (Event<?> event : events) {
|
---|
| 332 | for (List<? extends Event<?>> seqShorter : seqsShorter) {
|
---|
[94] | 333 | Event<?> lastEvent = event;
|
---|
[100] | 334 | if (getProbability(seqShorter, lastEvent) > 0.0) {
|
---|
[94] | 335 | List<Event<?>> subSeq = new ArrayList<Event<?>>(seqShorter);
|
---|
| 336 | subSeq.add(lastEvent);
|
---|
| 337 | sequenceSet.add(subSeq);
|
---|
| 338 | }
|
---|
| 339 | }
|
---|
| 340 | }
|
---|
| 341 | return sequenceSet;
|
---|
| 342 | }
|
---|
[100] | 343 |
|
---|
| 344 | /*
|
---|
| 345 | * (non-Javadoc)
|
---|
| 346 | *
|
---|
| 347 | * @see
|
---|
| 348 | * de.ugoe.cs.eventbench.models.IStochasticProcess#generateValidSequences
|
---|
| 349 | * (int)
|
---|
| 350 | */
|
---|
| 351 | @Override
|
---|
[118] | 352 | public Collection<List<? extends Event<?>>> generateValidSequences(
|
---|
| 353 | int length) {
|
---|
[94] | 354 | // check for min-length implicitly done by generateSequences
|
---|
[118] | 355 | Collection<List<? extends Event<?>>> allSequences = generateSequences(
|
---|
| 356 | length, true);
|
---|
[102] | 357 | Collection<List<? extends Event<?>>> validSequences = new LinkedHashSet<List<? extends Event<?>>>();
|
---|
[100] | 358 | for (List<? extends Event<?>> sequence : allSequences) {
|
---|
| 359 | if (sequence.size() == length
|
---|
| 360 | && Event.ENDEVENT.equals(sequence.get(sequence.size() - 1))) {
|
---|
[95] | 361 | validSequences.add(sequence);
|
---|
[94] | 362 | }
|
---|
| 363 | }
|
---|
| 364 | return validSequences;
|
---|
| 365 | }
|
---|
[12] | 366 |
|
---|
[118] | 367 | /*
|
---|
| 368 | * (non-Javadoc)
|
---|
| 369 | *
|
---|
| 370 | * @see
|
---|
| 371 | * de.ugoe.cs.eventbench.models.IStochasticProcess#getProbability(java.util
|
---|
| 372 | * .List)
|
---|
| 373 | */
|
---|
| 374 | @Override
|
---|
| 375 | public double getProbability(List<? extends Event<?>> sequence) {
|
---|
[342] | 376 | if (sequence == null) {
|
---|
| 377 | throw new InvalidParameterException("sequence must not be null");
|
---|
| 378 | }
|
---|
[118] | 379 | double prob = 1.0;
|
---|
[342] | 380 | List<Event<?>> context = new LinkedList<Event<?>>();
|
---|
| 381 | for (Event<?> event : sequence) {
|
---|
| 382 | prob *= getProbability(context, event);
|
---|
| 383 | context.add(event);
|
---|
[118] | 384 | }
|
---|
| 385 | return prob;
|
---|
| 386 | }
|
---|
| 387 |
|
---|
[182] | 388 | /*
|
---|
| 389 | * (non-Javadoc)
|
---|
| 390 | *
|
---|
[129] | 391 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#getNumFOMStates()
|
---|
| 392 | */
|
---|
| 393 | @Override
|
---|
| 394 | public int getNumFOMStates() {
|
---|
[252] | 395 | if (trie == null) {
|
---|
| 396 | return 0;
|
---|
| 397 | } else {
|
---|
| 398 | return trie.getNumLeafAncestors();
|
---|
| 399 | }
|
---|
[129] | 400 | }
|
---|
[248] | 401 |
|
---|
| 402 | /*
|
---|
| 403 | * (non-Javadoc)
|
---|
| 404 | *
|
---|
| 405 | * @see de.ugoe.cs.eventbench.models.IStochasticProcess#getNumTransitions()
|
---|
| 406 | */
|
---|
| 407 | @Override
|
---|
| 408 | public int getNumTransitions() {
|
---|
[252] | 409 | if (trie == null) {
|
---|
| 410 | return 0;
|
---|
| 411 | } else {
|
---|
| 412 | return trie.getNumLeafs();
|
---|
| 413 | }
|
---|
[248] | 414 | }
|
---|
[12] | 415 | } |
---|