001 /**
002 * Copyright (C) 2007-2008, Jens Lehmann
003 *
004 * This file is part of DL-Learner.
005 *
006 * DL-Learner is free software; you can redistribute it and/or modify
007 * it under the terms of the GNU General Public License as published by
008 * the Free Software Foundation; either version 3 of the License, or
009 * (at your option) any later version.
010 *
011 * DL-Learner is distributed in the hope that it will be useful,
012 * but WITHOUT ANY WARRANTY; without even the implied warranty of
013 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
014 * GNU General Public License for more details.
015 *
016 * You should have received a copy of the GNU General Public License
017 * along with this program. If not, see <http://www.gnu.org/licenses/>.
018 *
019 */
020
021 package org.dllearner.algorithms.refinement2;
022
023 import org.dllearner.utilities.owl.ConceptComparator;
024
025 /**
026 * This heuristic compares two nodes by computing a score
027 * using the number of covered negatives and the horizontal
028 * expansion factor of a node as input. Using this score
029 * it decides which one of the nodes seems to be more promising.
030 * The heuristic is flexible, because it offers a tradeoff
031 * between accurary and horizontal expansion (concept length).
032 * In contrast to the lexicographic heuristic this means that
033 * it sometimes prefers worse classifiers with low horizontal
034 * expansion over a better classifier with high horizontal
035 * expansion.
036 *
037 * It can be configured by using the "percentPerLenghtUnit"
038 * constructor argument. A higher
039 * value means that the algorithm is more likely to search in
040 * unexplored areas (= low horizontal expansion) of the search
041 * space vs. looking in promising but already explored (= high
042 * horizontal expansion) areas of the search space.
043 *
044 * @author Jens Lehmann
045 *
046 */
047 public class FlexibleHeuristic implements ExampleBasedHeuristic {
048
049 // Vergleich von Konzepten, falls alle anderen Kriterien fehlschlagen
050 private ConceptComparator conceptComparator = new ConceptComparator();
051 private int nrOfNegativeExamples;
052 private double percentPerLengthUnit;
053
054 // 5% sind eine Verlängerung um 1 wert
055 // double percentPerLengthUnit = 0.05;
056
057 public FlexibleHeuristic(int nrOfNegativeExamples, double percentPerLengthUnit) {
058 this.nrOfNegativeExamples = nrOfNegativeExamples;
059 this.percentPerLengthUnit = percentPerLengthUnit;
060 }
061
062 // implementiert einfach die Definition in der Diplomarbeit
063 public int compare(ExampleBasedNode n1, ExampleBasedNode n2) {
064
065 // sicherstellen, dass Qualität ausgewertet wurde
066 if(n1.isQualityEvaluated() && n2.isQualityEvaluated() && !n1.isTooWeak() && !n2.isTooWeak()) {
067
068 // alle scores sind negativ, größere scores sind besser
069 double score1 = -n1.getCoveredNegatives().size()/(double)nrOfNegativeExamples;
070 score1 -= percentPerLengthUnit * n1.getConcept().getLength();
071
072 double score2 = -n2.getCoveredNegatives().size()/(double)nrOfNegativeExamples;
073 score2 -= percentPerLengthUnit * n2.getConcept().getLength();
074
075 double diff = score1 - score2;
076
077 if(diff>0)
078 return 1;
079 else if(diff<0)
080 return -1;
081 else
082 return conceptComparator.compare(n1.getConcept(), n2.getConcept());
083 }
084
085 throw new RuntimeException("Cannot compare nodes, which have no evaluated quality or are too weak.");
086 }
087
088 @Override
089 public boolean equals(Object o) {
090 return (o instanceof FlexibleHeuristic);
091 }
092
093 }