annotate src/pagerank/LinkToVertex.java @ 2:1744340f8be6 draft

add some java files
author one
date Wed, 05 Sep 2012 11:56:21 +0900
parents
children dcd59917a2dd
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
1744340f8be6 add some java files
one
parents:
diff changeset
1 package pagerank;
1744340f8be6 add some java files
one
parents:
diff changeset
2
1744340f8be6 add some java files
one
parents:
diff changeset
3 import java.util.HashMap;
1744340f8be6 add some java files
one
parents:
diff changeset
4
1744340f8be6 add some java files
one
parents:
diff changeset
5 import com.tinkerpop.blueprints.Direction;
1744340f8be6 add some java files
one
parents:
diff changeset
6 import com.tinkerpop.blueprints.Edge;
1744340f8be6 add some java files
one
parents:
diff changeset
7 import com.tinkerpop.blueprints.Graph;
1744340f8be6 add some java files
one
parents:
diff changeset
8 import com.tinkerpop.blueprints.Vertex;
1744340f8be6 add some java files
one
parents:
diff changeset
9
1744340f8be6 add some java files
one
parents:
diff changeset
10 import pagerank.WikiPage;
1744340f8be6 add some java files
one
parents:
diff changeset
11
1744340f8be6 add some java files
one
parents:
diff changeset
12 public class LinkToVertex {
1744340f8be6 add some java files
one
parents:
diff changeset
13
1744340f8be6 add some java files
one
parents:
diff changeset
14 Graph graph;
1744340f8be6 add some java files
one
parents:
diff changeset
15 public final static String PAGE_TITLE = "pageTitle";
1744340f8be6 add some java files
one
parents:
diff changeset
16 public final static String PAGE_RANK = "pageRank";
1744340f8be6 add some java files
one
parents:
diff changeset
17 private HashMap<String, Long> pageIdTable = new HashMap<String, Long>();
1744340f8be6 add some java files
one
parents:
diff changeset
18
1744340f8be6 add some java files
one
parents:
diff changeset
19 private HashMap<String, WikiPage> wikiPageHash = new HashMap<String, WikiPage>();
1744340f8be6 add some java files
one
parents:
diff changeset
20 private long AllNodeNumber;
1744340f8be6 add some java files
one
parents:
diff changeset
21
1744340f8be6 add some java files
one
parents:
diff changeset
22 private final double weight1 = 0.85;
1744340f8be6 add some java files
one
parents:
diff changeset
23 private final double weight2 = 0.15;
1744340f8be6 add some java files
one
parents:
diff changeset
24
1744340f8be6 add some java files
one
parents:
diff changeset
25 public static final String HAS_LINK = "HasLink";
1744340f8be6 add some java files
one
parents:
diff changeset
26
1744340f8be6 add some java files
one
parents:
diff changeset
27 LinkToVertex(Graph graph) {
1744340f8be6 add some java files
one
parents:
diff changeset
28 this.graph = graph;
1744340f8be6 add some java files
one
parents:
diff changeset
29 AllNodeNumber = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
30 }
1744340f8be6 add some java files
one
parents:
diff changeset
31
1744340f8be6 add some java files
one
parents:
diff changeset
32 Long getId(String pageTitle) {
1744340f8be6 add some java files
one
parents:
diff changeset
33 return pageIdTable.get(pageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
34 }
1744340f8be6 add some java files
one
parents:
diff changeset
35
1744340f8be6 add some java files
one
parents:
diff changeset
36 boolean isHasLink(String label) {
1744340f8be6 add some java files
one
parents:
diff changeset
37 return label.equals(HAS_LINK);
1744340f8be6 add some java files
one
parents:
diff changeset
38 }
1744340f8be6 add some java files
one
parents:
diff changeset
39
1744340f8be6 add some java files
one
parents:
diff changeset
40 private Vertex createVertex() {
1744340f8be6 add some java files
one
parents:
diff changeset
41 return graph.addVertex(null);
1744340f8be6 add some java files
one
parents:
diff changeset
42 }
1744340f8be6 add some java files
one
parents:
diff changeset
43
1744340f8be6 add some java files
one
parents:
diff changeset
44 private Vertex createVertex(Object id) {
1744340f8be6 add some java files
one
parents:
diff changeset
45 return graph.addVertex(id);
1744340f8be6 add some java files
one
parents:
diff changeset
46 }
1744340f8be6 add some java files
one
parents:
diff changeset
47
1744340f8be6 add some java files
one
parents:
diff changeset
48 private Vertex createVertexWithProperty(String key, Object value) {
1744340f8be6 add some java files
one
parents:
diff changeset
49 Vertex v = graph.addVertex(null);
1744340f8be6 add some java files
one
parents:
diff changeset
50 v.setProperty(key,value);
1744340f8be6 add some java files
one
parents:
diff changeset
51 return v;
1744340f8be6 add some java files
one
parents:
diff changeset
52 }
1744340f8be6 add some java files
one
parents:
diff changeset
53
1744340f8be6 add some java files
one
parents:
diff changeset
54 String getPageTitle(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
55 return (String) v.getProperty(PAGE_TITLE);
1744340f8be6 add some java files
one
parents:
diff changeset
56 }
1744340f8be6 add some java files
one
parents:
diff changeset
57
1744340f8be6 add some java files
one
parents:
diff changeset
58 Double getPageRank(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
59 return (Double) v.getProperty(PAGE_RANK);
1744340f8be6 add some java files
one
parents:
diff changeset
60 }
1744340f8be6 add some java files
one
parents:
diff changeset
61
1744340f8be6 add some java files
one
parents:
diff changeset
62 Vertex createVertexWithPageTitle(String pageTitle) {
1744340f8be6 add some java files
one
parents:
diff changeset
63 Vertex v = createVertexWithProperty(PAGE_TITLE, pageTitle);
1744340f8be6 add some java files
one
parents:
diff changeset
64 pageIdTable.put(pageTitle, (Long) v.getId());
1744340f8be6 add some java files
one
parents:
diff changeset
65 return v;
1744340f8be6 add some java files
one
parents:
diff changeset
66 }
1744340f8be6 add some java files
one
parents:
diff changeset
67
1744340f8be6 add some java files
one
parents:
diff changeset
68 Vertex setPageRank(Vertex v, Double rank) {
1744340f8be6 add some java files
one
parents:
diff changeset
69 v.setProperty(PAGE_RANK, rank);
1744340f8be6 add some java files
one
parents:
diff changeset
70 return v;
1744340f8be6 add some java files
one
parents:
diff changeset
71 }
1744340f8be6 add some java files
one
parents:
diff changeset
72
1744340f8be6 add some java files
one
parents:
diff changeset
73 Vertex getVertex(String name) {
1744340f8be6 add some java files
one
parents:
diff changeset
74 long id = pageIdTable.get(name);
1744340f8be6 add some java files
one
parents:
diff changeset
75 return graph.getVertex(id);
1744340f8be6 add some java files
one
parents:
diff changeset
76 }
1744340f8be6 add some java files
one
parents:
diff changeset
77
1744340f8be6 add some java files
one
parents:
diff changeset
78 Vertex getNode(int nodeId) {
1744340f8be6 add some java files
one
parents:
diff changeset
79 return graph.getVertex(nodeId);
1744340f8be6 add some java files
one
parents:
diff changeset
80 }
1744340f8be6 add some java files
one
parents:
diff changeset
81
1744340f8be6 add some java files
one
parents:
diff changeset
82 Edge setRelationship(Vertex v1, Vertex v2, String label) {
1744340f8be6 add some java files
one
parents:
diff changeset
83 Edge e = graph.addEdge(null, v1, v2, label);
1744340f8be6 add some java files
one
parents:
diff changeset
84 return e;
1744340f8be6 add some java files
one
parents:
diff changeset
85 }
1744340f8be6 add some java files
one
parents:
diff changeset
86
1744340f8be6 add some java files
one
parents:
diff changeset
87 Edge setHasLink(Vertex v1, Vertex v2) {
1744340f8be6 add some java files
one
parents:
diff changeset
88 return setRelationship(v1, v2, HAS_LINK);
1744340f8be6 add some java files
one
parents:
diff changeset
89 }
1744340f8be6 add some java files
one
parents:
diff changeset
90
1744340f8be6 add some java files
one
parents:
diff changeset
91 long searchAllNodes() {
1744340f8be6 add some java files
one
parents:
diff changeset
92 AllNodeNumber = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
93 for (Vertex v : graph.getVertices()) {
1744340f8be6 add some java files
one
parents:
diff changeset
94 if ( (v.getProperty(PAGE_TITLE) != null) &&
1744340f8be6 add some java files
one
parents:
diff changeset
95 (v.getProperty(PAGE_RANK)) != null ) {
1744340f8be6 add some java files
one
parents:
diff changeset
96 WikiPage wiki = new WikiPage(v);
1744340f8be6 add some java files
one
parents:
diff changeset
97 pageIdTable.put((String) v.getProperty(PAGE_TITLE), (Long) v.getId());
1744340f8be6 add some java files
one
parents:
diff changeset
98 wiki.setInHasLink(computeInHasLink(v));
1744340f8be6 add some java files
one
parents:
diff changeset
99 wiki.setOutHasLink(computeOutHasLink(v));
1744340f8be6 add some java files
one
parents:
diff changeset
100 wikiPageHash.put((String) v.getProperty(PAGE_TITLE), wiki);
1744340f8be6 add some java files
one
parents:
diff changeset
101 AllNodeNumber++;
1744340f8be6 add some java files
one
parents:
diff changeset
102 }
1744340f8be6 add some java files
one
parents:
diff changeset
103 }
1744340f8be6 add some java files
one
parents:
diff changeset
104 return AllNodeNumber;
1744340f8be6 add some java files
one
parents:
diff changeset
105 }
1744340f8be6 add some java files
one
parents:
diff changeset
106
1744340f8be6 add some java files
one
parents:
diff changeset
107 void searchRegiNodes(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
108
1744340f8be6 add some java files
one
parents:
diff changeset
109 if ( (v.getProperty(PAGE_TITLE) != null) &&
1744340f8be6 add some java files
one
parents:
diff changeset
110 (v.getProperty(PAGE_RANK) != null)) {
1744340f8be6 add some java files
one
parents:
diff changeset
111 WikiPage wiki = new WikiPage(v);
1744340f8be6 add some java files
one
parents:
diff changeset
112 pageIdTable.put((String) v.getProperty(PAGE_TITLE), (Long) v.getId());
1744340f8be6 add some java files
one
parents:
diff changeset
113 wiki.setInHasLink(computeInHasLink(v));
1744340f8be6 add some java files
one
parents:
diff changeset
114 wiki.setOutHasLink(computeOutHasLink(v));
1744340f8be6 add some java files
one
parents:
diff changeset
115 wikiPageHash.put((String) v.getProperty(PAGE_TITLE), wiki);
1744340f8be6 add some java files
one
parents:
diff changeset
116 AllNodeNumber++;
1744340f8be6 add some java files
one
parents:
diff changeset
117 }
1744340f8be6 add some java files
one
parents:
diff changeset
118 }
1744340f8be6 add some java files
one
parents:
diff changeset
119
1744340f8be6 add some java files
one
parents:
diff changeset
120 HashMap<String, WikiPage> getWikiPageHash() {
1744340f8be6 add some java files
one
parents:
diff changeset
121 return wikiPageHash;
1744340f8be6 add some java files
one
parents:
diff changeset
122 }
1744340f8be6 add some java files
one
parents:
diff changeset
123
1744340f8be6 add some java files
one
parents:
diff changeset
124 HashMap<String, Long> getPageIdTable() {
1744340f8be6 add some java files
one
parents:
diff changeset
125 return pageIdTable;
1744340f8be6 add some java files
one
parents:
diff changeset
126 }
1744340f8be6 add some java files
one
parents:
diff changeset
127
1744340f8be6 add some java files
one
parents:
diff changeset
128 public Iterable<Vertex> getAllNodes() {
1744340f8be6 add some java files
one
parents:
diff changeset
129 return graph.getVertices();
1744340f8be6 add some java files
one
parents:
diff changeset
130 }
1744340f8be6 add some java files
one
parents:
diff changeset
131
1744340f8be6 add some java files
one
parents:
diff changeset
132 public void printAllNodes() {
1744340f8be6 add some java files
one
parents:
diff changeset
133 for (Vertex v : graph.getVertices() ) {
1744340f8be6 add some java files
one
parents:
diff changeset
134 System.out.println("ID = "+ v.getId());
1744340f8be6 add some java files
one
parents:
diff changeset
135 for (String key: v.getPropertyKeys()) {
1744340f8be6 add some java files
one
parents:
diff changeset
136 System.out.println(key + "=" + v.getProperty(key));
1744340f8be6 add some java files
one
parents:
diff changeset
137 }
1744340f8be6 add some java files
one
parents:
diff changeset
138 /*
1744340f8be6 add some java files
one
parents:
diff changeset
139 for (Edge e : v.getEdges(Direction.IN, HAS_LINK) ) {
1744340f8be6 add some java files
one
parents:
diff changeset
140 System.out.println();
1744340f8be6 add some java files
one
parents:
diff changeset
141 }
1744340f8be6 add some java files
one
parents:
diff changeset
142 */
1744340f8be6 add some java files
one
parents:
diff changeset
143 }
1744340f8be6 add some java files
one
parents:
diff changeset
144 System.out.println("--");
1744340f8be6 add some java files
one
parents:
diff changeset
145
1744340f8be6 add some java files
one
parents:
diff changeset
146 }
1744340f8be6 add some java files
one
parents:
diff changeset
147
1744340f8be6 add some java files
one
parents:
diff changeset
148 public long computeOutHasLink(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
149 long count = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
150 for (Edge edge : v.getEdges(Direction.OUT, HAS_LINK)) {
1744340f8be6 add some java files
one
parents:
diff changeset
151 count++;
1744340f8be6 add some java files
one
parents:
diff changeset
152 }
1744340f8be6 add some java files
one
parents:
diff changeset
153 return count;
1744340f8be6 add some java files
one
parents:
diff changeset
154 }
1744340f8be6 add some java files
one
parents:
diff changeset
155
1744340f8be6 add some java files
one
parents:
diff changeset
156 public long computeInHasLink(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
157 long count = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
158 for (Edge edge : v.getEdges(Direction.IN, HAS_LINK)) {
1744340f8be6 add some java files
one
parents:
diff changeset
159 count++;
1744340f8be6 add some java files
one
parents:
diff changeset
160 }
1744340f8be6 add some java files
one
parents:
diff changeset
161 return count;
1744340f8be6 add some java files
one
parents:
diff changeset
162 }
1744340f8be6 add some java files
one
parents:
diff changeset
163
1744340f8be6 add some java files
one
parents:
diff changeset
164 public void printOutHasLink(Vertex v, int depth) {
1744340f8be6 add some java files
one
parents:
diff changeset
165 int numberOfLinkPages = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
166 String output = v.getProperty(PAGE_TITLE) + " outHasLink pages:";
1744340f8be6 add some java files
one
parents:
diff changeset
167 System.out.println(output);
1744340f8be6 add some java files
one
parents:
diff changeset
168 for (Edge edge : v.getEdges(Direction.OUT, HAS_LINK)) {
1744340f8be6 add some java files
one
parents:
diff changeset
169 Vertex outV = edge.getVertex(Direction.IN);
1744340f8be6 add some java files
one
parents:
diff changeset
170 String str = (String) outV.getProperty(PAGE_TITLE);
1744340f8be6 add some java files
one
parents:
diff changeset
171 System.out.println(str);
1744340f8be6 add some java files
one
parents:
diff changeset
172 numberOfLinkPages++;
1744340f8be6 add some java files
one
parents:
diff changeset
173 }
1744340f8be6 add some java files
one
parents:
diff changeset
174 String numOutput = "Number of outHaslink pages: " + numberOfLinkPages;
1744340f8be6 add some java files
one
parents:
diff changeset
175 System.out.println(numOutput);
1744340f8be6 add some java files
one
parents:
diff changeset
176 }
1744340f8be6 add some java files
one
parents:
diff changeset
177
1744340f8be6 add some java files
one
parents:
diff changeset
178 public void printInHasLink(Vertex v, int depth) {
1744340f8be6 add some java files
one
parents:
diff changeset
179 int numberOfLinkPages = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
180 String output = v.getProperty(PAGE_TITLE) + " inHasLink pages:";
1744340f8be6 add some java files
one
parents:
diff changeset
181 System.out.println(output);
1744340f8be6 add some java files
one
parents:
diff changeset
182 for (Edge edge : v.getEdges(Direction.IN, HAS_LINK)) {
1744340f8be6 add some java files
one
parents:
diff changeset
183 Vertex outV = edge.getVertex(Direction.OUT);
1744340f8be6 add some java files
one
parents:
diff changeset
184 String str = (String) outV.getProperty(PAGE_TITLE);
1744340f8be6 add some java files
one
parents:
diff changeset
185 System.out.println(str);
1744340f8be6 add some java files
one
parents:
diff changeset
186 numberOfLinkPages++;
1744340f8be6 add some java files
one
parents:
diff changeset
187 }
1744340f8be6 add some java files
one
parents:
diff changeset
188 String numOutput = "Number of inHaslink pages: " + numberOfLinkPages + "\n";
1744340f8be6 add some java files
one
parents:
diff changeset
189 System.out.println(numOutput);
1744340f8be6 add some java files
one
parents:
diff changeset
190 }
1744340f8be6 add some java files
one
parents:
diff changeset
191
1744340f8be6 add some java files
one
parents:
diff changeset
192 public double computePageRank(Vertex v) {
1744340f8be6 add some java files
one
parents:
diff changeset
193 double sum = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
194 double pageRank = 0;
1744340f8be6 add some java files
one
parents:
diff changeset
195 String title = getPageTitle(v);
1744340f8be6 add some java files
one
parents:
diff changeset
196 WikiPage wiki = wikiPageHash.get(title);
1744340f8be6 add some java files
one
parents:
diff changeset
197
1744340f8be6 add some java files
one
parents:
diff changeset
198 for (Edge edge : v.getEdges(Direction.IN, HAS_LINK) ) {
1744340f8be6 add some java files
one
parents:
diff changeset
199 Vertex linkV = edge.getVertex(Direction.OUT);
1744340f8be6 add some java files
one
parents:
diff changeset
200 sum += (double) ((Double) linkV.getProperty(PAGE_RANK)) / computeInHasLink(linkV) ;
1744340f8be6 add some java files
one
parents:
diff changeset
201 }
1744340f8be6 add some java files
one
parents:
diff changeset
202
1744340f8be6 add some java files
one
parents:
diff changeset
203 if (computeOutHasLink(v) == 0) {
1744340f8be6 add some java files
one
parents:
diff changeset
204 pageRank = (double) sum * weight1
1744340f8be6 add some java files
one
parents:
diff changeset
205 + (double) ((double) 1 / AllNodeNumber * weight2);
1744340f8be6 add some java files
one
parents:
diff changeset
206 } else {
1744340f8be6 add some java files
one
parents:
diff changeset
207 pageRank = (double) ((double)sum / computeOutHasLink(v) * weight1)
1744340f8be6 add some java files
one
parents:
diff changeset
208 + (double) ((double) 1 / AllNodeNumber * weight2);
1744340f8be6 add some java files
one
parents:
diff changeset
209 }
1744340f8be6 add some java files
one
parents:
diff changeset
210 wiki.setRank(pageRank);
1744340f8be6 add some java files
one
parents:
diff changeset
211 v.setProperty(PAGE_RANK, pageRank);
1744340f8be6 add some java files
one
parents:
diff changeset
212 return pageRank;
1744340f8be6 add some java files
one
parents:
diff changeset
213 }
1744340f8be6 add some java files
one
parents:
diff changeset
214
1744340f8be6 add some java files
one
parents:
diff changeset
215 public void printNodeInfo(int nodeId) {
1744340f8be6 add some java files
one
parents:
diff changeset
216 Vertex v = graph.getVertex(nodeId);
1744340f8be6 add some java files
one
parents:
diff changeset
217 printInHasLink(v, 1);
1744340f8be6 add some java files
one
parents:
diff changeset
218 printOutHasLink(v, 1);
1744340f8be6 add some java files
one
parents:
diff changeset
219
1744340f8be6 add some java files
one
parents:
diff changeset
220 String title = getPageTitle(v);
1744340f8be6 add some java files
one
parents:
diff changeset
221 double rank = getPageRank(v);
1744340f8be6 add some java files
one
parents:
diff changeset
222 long inHasLink = computeInHasLink(v);
1744340f8be6 add some java files
one
parents:
diff changeset
223 long outHasLink = computeOutHasLink(v);
1744340f8be6 add some java files
one
parents:
diff changeset
224
1744340f8be6 add some java files
one
parents:
diff changeset
225 System.out.println("id:"+nodeId+" title:"+title+" rank:"+rank);
1744340f8be6 add some java files
one
parents:
diff changeset
226 System.out.println("inHasLink:"+inHasLink+" outHasLink:"+outHasLink);
1744340f8be6 add some java files
one
parents:
diff changeset
227
1744340f8be6 add some java files
one
parents:
diff changeset
228
1744340f8be6 add some java files
one
parents:
diff changeset
229 }
1744340f8be6 add some java files
one
parents:
diff changeset
230
1744340f8be6 add some java files
one
parents:
diff changeset
231 }