Line | Hits | Source |
---|---|---|
1 | /* | |
2 | * Copyright (c) 2003, the JUNG Project and the Regents of the University | |
3 | * of California | |
4 | * All rights reserved. | |
5 | * | |
6 | * This software is open-source under the BSD license; see either | |
7 | * "license.txt" or | |
8 | * http://jung.sourceforge.net/license.txt for a description. | |
9 | */ | |
10 | package edu.uci.ics.jung.algorithms.importance; | |
11 | ||
12 | import java.util.Iterator; | |
13 | import java.util.List; | |
14 | import java.util.Set; | |
15 | ||
16 | import cern.colt.matrix.DoubleMatrix1D; | |
17 | import cern.colt.matrix.DoubleMatrix2D; | |
18 | import cern.colt.matrix.impl.DenseDoubleMatrix1D; | |
19 | import cern.colt.matrix.impl.SparseDoubleMatrix1D; | |
20 | import edu.uci.ics.jung.algorithms.GraphMatrixOperations; | |
21 | import edu.uci.ics.jung.graph.ArchetypeVertex; | |
22 | import edu.uci.ics.jung.graph.DirectedGraph; | |
23 | import edu.uci.ics.jung.graph.Element; | |
24 | import edu.uci.ics.jung.graph.Vertex; | |
25 | import edu.uci.ics.jung.graph.decorators.Indexer; | |
26 | import edu.uci.ics.jung.utils.MutableDouble; | |
27 | import edu.uci.ics.jung.utils.UserData; | |
28 | ||
29 | /** | |
30 | * @author Scott White and Joshua O'Madadhain | |
31 | * @see "Algorithms for Estimating Relative Importance in Graphs by Scott White and Padhraic Smyth, 2003" | |
32 | */ | |
33 | public class MarkovCentrality extends RelativeAuthorityRanker { | |
34 | public final static String MEAN_FIRST_PASSAGE_TIME = "jung.algorithms.importance.mean_first_passage_time"; | |
35 | private DoubleMatrix1D mRankings; | |
36 | private Indexer mIndexer; | |
37 | ||
38 | public MarkovCentrality(DirectedGraph graph, Set rootNodes) { | |
39 | 0 | this(graph,rootNodes,null); |
40 | 0 | } |
41 | ||
42 | 1 | public MarkovCentrality(DirectedGraph graph, Set rootNodes, String edgeWeightKey) { |
43 | 1 | super.initialize(graph, true, false); |
44 | 1 | setPriors(rootNodes); |
45 | 1 | if (edgeWeightKey == null) |
46 | 1 | assignDefaultEdgeTransitionWeights(); |
47 | else | |
48 | 0 | setUserDefinedEdgeWeightKey(edgeWeightKey); |
49 | 1 | normalizeEdgeTransitionWeights(); |
50 | ||
51 | 1 | mIndexer = Indexer.getIndexer(graph); |
52 | 1 | mRankings = new SparseDoubleMatrix1D(graph.numVertices()); |
53 | 1 | } |
54 | ||
55 | /** | |
56 | * @see edu.uci.ics.jung.algorithms.importance.AbstractRanker#getRankScoreKey() | |
57 | */ | |
58 | public String getRankScoreKey() { | |
59 | 0 | return MEAN_FIRST_PASSAGE_TIME; |
60 | } | |
61 | ||
62 | /** | |
63 | * @see edu.uci.ics.jung.algorithms.importance.AbstractRanker#getRankScore(edu.uci.ics.jung.graph.Element) | |
64 | */ | |
65 | public double getRankScore(Element vert) { | |
66 | 8 | ArchetypeVertex v = (ArchetypeVertex) vert; |
67 | 8 | return mRankings.get(mIndexer.getIndex(v)); |
68 | } | |
69 | ||
70 | /** | |
71 | * @see edu.uci.ics.jung.algorithms.importance.AbstractRanker#setRankScore(edu.uci.ics.jung.graph.Element, double) | |
72 | */ | |
73 | protected void setRankScore(Element v, double rankValue) { | |
74 | 0 | v.setUserDatum(getRankScoreKey(), new MutableDouble(rankValue), UserData.SHARED); |
75 | 0 | } |
76 | ||
77 | /** | |
78 | * @see edu.uci.ics.jung.algorithms.IterativeProcess#evaluateIteration() | |
79 | */ | |
80 | protected double evaluateIteration() { | |
81 | 1 | DoubleMatrix2D mFPTMatrix = GraphMatrixOperations.computeMeanFirstPassageMatrix(getGraph(), getEdgeWeightKeyName(), getStationaryDistribution()); |
82 | ||
83 | 1 | mRankings.assign(0); |
84 | ||
85 | 1 | for (Iterator p_iter = getPriors().iterator(); p_iter.hasNext();) { |
86 | 1 | Vertex p = (Vertex) p_iter.next(); |
87 | 1 | int p_id = mIndexer.getIndex(p); |
88 | 1 | for (Iterator v_iter = getVertices().iterator(); v_iter.hasNext();) { |
89 | 4 | Vertex v = (Vertex) v_iter.next(); |
90 | 4 | int v_id = mIndexer.getIndex(v); |
91 | 4 | mRankings.set(v_id, mRankings.get(v_id) + mFPTMatrix.get(p_id, v_id)); |
92 | } | |
93 | } | |
94 | ||
95 | 1 | for (Iterator v_iter = getVertices().iterator(); v_iter.hasNext();) { |
96 | 4 | Vertex v = (Vertex) v_iter.next(); |
97 | 4 | int v_id = mIndexer.getIndex(v); |
98 | 4 | mRankings.set(v_id, 1 / (mRankings.get(v_id) / getPriors().size())); |
99 | } | |
100 | ||
101 | 1 | double total = mRankings.zSum(); |
102 | ||
103 | 1 | for (Iterator v_iter = getVertices().iterator(); v_iter.hasNext();) { |
104 | 4 | Vertex v = (Vertex) v_iter.next(); |
105 | 4 | int v_id = mIndexer.getIndex(v); |
106 | 4 | mRankings.set(v_id, mRankings.get(v_id) / total); |
107 | } | |
108 | ||
109 | 1 | return 0; |
110 | } | |
111 | ||
112 | ||
113 | /** | |
114 | * Loads the stationary distribution into a vector if it was passed in, | |
115 | * or calculates it if not. | |
116 | * | |
117 | * @return DoubleMatrix1D | |
118 | */ | |
119 | private DoubleMatrix1D getStationaryDistribution() { | |
120 | 1 | DoubleMatrix1D piVector = new DenseDoubleMatrix1D(getVertices().size()); |
121 | 1 | PageRank pageRank = new PageRank((DirectedGraph) getGraph(), 0, getEdgeWeightKeyName()); |
122 | 1 | pageRank.evaluate(); |
123 | 1 | List rankings = pageRank.getRankings(); |
124 | ||
125 | 1 | for (Iterator r_iter = rankings.iterator(); r_iter.hasNext();) { |
126 | 4 | NodeRanking rank = (NodeRanking) r_iter.next(); |
127 | 4 | piVector.set(mIndexer.getIndex(rank.vertex), rank.rankScore); |
128 | } | |
129 | 1 | return piVector; |
130 | } | |
131 | ||
132 | } |
this report was generated by version 1.0.5 of jcoverage. |
copyright © 2003, jcoverage ltd. all rights reserved. |