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