├── .DS_Store
├── .gitattributes
├── DBLPPreProcess
    ├── .DS_Store
    ├── .classpath
    ├── .project
    ├── .settings
    │   └── org.eclipse.jdt.core.prefs
    ├── bin
    │   └── .DS_Store
    └── src
    │   ├── PreProcessStep1.java
    │   ├── PreProcessStep2.java
    │   └── PreProcessStep3.java
├── README.md
└── RandomWalk2Vec
    ├── RandomWalk2VecRun.sh
    ├── compile.sh
    └── randomWalk2vec.c


/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/daokunzhang/MetaGraph2Vec/7fe0fab3d0948aca69e81b5d4e230927d75676d8/.DS_Store


--------------------------------------------------------------------------------
/.gitattributes:
--------------------------------------------------------------------------------
1 | # Auto detect text files and perform LF normalization
2 | * text=auto
3 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/daokunzhang/MetaGraph2Vec/7fe0fab3d0948aca69e81b5d4e230927d75676d8/DBLPPreProcess/.DS_Store


--------------------------------------------------------------------------------
/DBLPPreProcess/.classpath:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <classpath>
3 | 	<classpathentry kind="src" path="src"/>
4 | 	<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.8"/>
5 | 	<classpathentry kind="output" path="bin"/>
6 | </classpath>
7 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/.project:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0" encoding="UTF-8"?>
 2 | <projectDescription>
 3 | 	<name>DBLPPreProcess</name>
 4 | 	<comment></comment>
 5 | 	<projects>
 6 | 	</projects>
 7 | 	<buildSpec>
 8 | 		<buildCommand>
 9 | 			<name>org.eclipse.jdt.core.javabuilder</name>
10 | 			<arguments>
11 | 			</arguments>
12 | 		</buildCommand>
13 | 	</buildSpec>
14 | 	<natures>
15 | 		<nature>org.eclipse.jdt.core.javanature</nature>
16 | 	</natures>
17 | </projectDescription>
18 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/.settings/org.eclipse.jdt.core.prefs:
--------------------------------------------------------------------------------
 1 | eclipse.preferences.version=1
 2 | org.eclipse.jdt.core.compiler.codegen.inlineJsrBytecode=enabled
 3 | org.eclipse.jdt.core.compiler.codegen.targetPlatform=1.8
 4 | org.eclipse.jdt.core.compiler.codegen.unusedLocal=preserve
 5 | org.eclipse.jdt.core.compiler.compliance=1.8
 6 | org.eclipse.jdt.core.compiler.debug.lineNumber=generate
 7 | org.eclipse.jdt.core.compiler.debug.localVariable=generate
 8 | org.eclipse.jdt.core.compiler.debug.sourceFile=generate
 9 | org.eclipse.jdt.core.compiler.problem.assertIdentifier=error
10 | org.eclipse.jdt.core.compiler.problem.enumIdentifier=error
11 | org.eclipse.jdt.core.compiler.source=1.8
12 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/bin/.DS_Store:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/daokunzhang/MetaGraph2Vec/7fe0fab3d0948aca69e81b5d4e230927d75676d8/DBLPPreProcess/bin/.DS_Store


--------------------------------------------------------------------------------
/DBLPPreProcess/src/PreProcessStep1.java:
--------------------------------------------------------------------------------
  1 | import java.io.BufferedReader;
  2 | import java.io.BufferedWriter;
  3 | import java.io.File;
  4 | import java.io.FileInputStream;
  5 | import java.io.FileOutputStream;
  6 | import java.io.IOException;
  7 | import java.io.InputStreamReader;
  8 | import java.io.OutputStreamWriter;
  9 | import java.util.ArrayList;
 10 | 
 11 | public class PreProcessStep1 { // extract subgraph form the DBLP-Citation-Network-V3.txt network
 12 | 	
 13 | 	public static int nonAlphabet(char ch)
 14 | 	{
 15 | 		if(ch>='A'&&ch<='Z')
 16 | 			return 0;
 17 | 		if(ch>='a'&&ch<='z')
 18 | 			return 0;
 19 | 		return 1;
 20 | 	}
 21 | 
 22 | 	public static int check(String venue)
 23 | 	{
 24 | 		String[] selectedVenues = {"SIGMOD", "ICDE", "VLDB", "EDBT",
 25 | 				"PODS", "ICDT", "DASFAA", "SSDBM", "CIKM", "KDD", 
 26 | 				"ICDM", "SDM", "PKDD", "PAKDD", "IJCAI", "AAAI", 
 27 | 				"NIPS", "ICML", "ECML", "ACML", "IJCNN", "UAI",
 28 | 				"ECAI", "COLT", "ACL", "KR", "CVPR", "ICCV",
 29 | 				"ECCV", "ACCV", "MM", "ICPR", "ICIP", "ICME"};
 30 | 		int res = 0;
 31 | 		for(int i=0;i<selectedVenues.length;i++)
 32 | 		{
 33 | 			int index = venue.indexOf(selectedVenues[i]);
 34 | 			if(index!=-1)
 35 | 			{
 36 | 				if(index==0)
 37 | 				{
 38 | 					if(index+selectedVenues[i].length()>=venue.length())
 39 | 						res = 1;
 40 | 					else
 41 | 					{
 42 | 						char endChar = venue.charAt(index+selectedVenues[i].length());
 43 | 						if(nonAlphabet(endChar)==1)
 44 | 							res = 1;
 45 | 					}
 46 | 				}
 47 | 				else
 48 | 				{
 49 | 					char startChar = venue.charAt(index-1);
 50 | 					if(nonAlphabet(startChar)==1)
 51 | 					{
 52 | 						if(index+selectedVenues[i].length()>=venue.length())
 53 | 							res = 1;
 54 | 						else
 55 | 						{
 56 | 							char endChar = venue.charAt(index+selectedVenues[i].length());
 57 | 							if(nonAlphabet(endChar)==1)
 58 | 								res = 1;
 59 | 						}
 60 | 					}
 61 | 				}
 62 | 			}
 63 | 			if(res==1)
 64 | 				break;
 65 | 		}
 66 | 		if(res==1)
 67 | 		{
 68 | 			if(venue.indexOf("SIGMOD")!=-1&&venue.indexOf("Data Mining")!=-1)
 69 | 				res = 0;
 70 | 		}
 71 | 		return res;
 72 | 	}
 73 | 	
 74 | 	public static void main(String[] args) throws IOException {
 75 | 		// TODO Auto-generated method stub
 76 | 		BufferedReader br = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-Citation-Network-V3.txt"), "UTF-8"));
 77 | 		BufferedWriter bwTitle = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.paper.title.txt"), "UTF-8"));
 78 | 		BufferedWriter bwAuthor = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.paper.author.txt"), "UTF-8"));
 79 | 		BufferedWriter bwVenue = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.paper.venue.txt"), "UTF-8"));
 80 | 		BufferedWriter bwRef = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.paper.ref.txt"), "UTF-8"));
 81 | 		int numOfNode = 0;
 82 | 		String line = null;
 83 | 		br.readLine();
 84 | 		String curIndex = "";
 85 | 		String curTitle = "";
 86 | 		String curAuthors = "";
 87 | 		String[] curAuthorList = null;
 88 | 		String curVenue = "";
 89 | 		ArrayList<String> curRef = new ArrayList<String>();
 90 | 		int writeStatus = 0;
 91 | 		int maxRefSize = 0;
 92 | 		String maxRefIndex = "";
 93 | 		while((line=br.readLine())!=null)
 94 | 		{
 95 | 			if(line.indexOf("#*")==0)
 96 | 				curTitle = line.substring(2).trim();
 97 | 			if(line.indexOf("#c")==0)
 98 | 			{
 99 | 				curVenue = line.substring(2).trim();
100 | 				curVenue = curVenue.replaceAll(" ", "");
101 | 				if(check(curVenue)==1)
102 | 					writeStatus = 1;
103 | 				else
104 | 					writeStatus = 0;
105 | 			}
106 | 			if(line.indexOf("#%")==0)
107 | 				curRef.add(line.substring(2).trim());
108 | 			if(line.indexOf("#@")==0)
109 | 			{
110 | 				curAuthors = line.substring(2).trim();
111 | 				curAuthorList = curAuthors.split(",");
112 | 			}
113 | 			if(line.indexOf("#index")==0) 
114 | 				curIndex = line.substring(6).trim();
115 | 			if(line.trim().length()==0)
116 | 			{
117 | 				if((!curTitle.equals(""))&&(!curVenue.equals(""))&&(!curAuthors.equals(""))&&(!curIndex.equals(""))&&writeStatus==1)
118 | 				{
119 | 					curIndex = curIndex.replaceAll(" ", "");
120 | 					bwTitle.write("p_"+curIndex+"\t"+curTitle+"\n");
121 | 					for(int i=0;i<curAuthorList.length;i++) {
122 | 						String curAuthor = curAuthorList[i];
123 | 						curAuthor = curAuthor.replaceAll(" ", "");
124 | 						if(!curAuthor.equals(""))
125 | 							bwAuthor.write("p_"+curIndex+"\ta_"+curAuthor+"\n");
126 | 					}
127 | 					bwVenue.write("p_"+curIndex+"\tv_"+curVenue+"\n");
128 | 					//bwRef.write("p_"+curIndex+"\tsize:\t"+curRef.size()+"\n");
129 | 					for(int i=0;i<curRef.size();i++)
130 | 					{
131 | 						String curReference = curRef.get(i);
132 | 						curReference = curReference.replaceAll(" ", "");
133 | 						if(!curReference.equals(""))
134 | 							bwRef.write("p_"+curIndex+"\tp_"+curReference+"\n");
135 | 					}
136 | 					if(curRef.size()>maxRefSize)
137 | 					{
138 | 						maxRefSize = curRef.size();
139 | 						maxRefIndex = curIndex;
140 | 					}
141 | 					numOfNode++;
142 | 				}
143 | 				curTitle = "";
144 | 				curAuthors = "";
145 | 				curVenue = "";
146 | 				curRef.clear();
147 | 				curIndex = "";
148 | 			}
149 | 		}
150 | 		br.close();
151 | 		bwTitle.close();
152 | 		bwAuthor.close();
153 | 		bwVenue.close();
154 | 		bwRef.close();
155 | 		System.out.println("numOfNode = "+numOfNode);
156 | 		System.out.println("maxRefSize = "+maxRefSize);
157 | 		System.out.println("maxRefIndex = "+maxRefIndex);
158 | 	}
159 | 
160 | }
161 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/src/PreProcessStep2.java:
--------------------------------------------------------------------------------
  1 | import java.io.BufferedReader;
  2 | import java.io.BufferedWriter;
  3 | import java.io.File;
  4 | import java.io.FileInputStream;
  5 | import java.io.FileOutputStream;
  6 | import java.io.IOException;
  7 | import java.io.InputStreamReader;
  8 | import java.io.OutputStreamWriter;
  9 | import java.util.ArrayList;
 10 | import java.util.Collections;
 11 | import java.util.Comparator;
 12 | import java.util.HashMap;
 13 | import java.util.Map;
 14 | import java.util.Random;
 15 | 
 16 | public class PreProcessStep2 { // obtain the class labels for authors;
 17 | 	
 18 | 	public static int getClass(String venue)
 19 | 	{
 20 | 		String[] venues1 = {"SIGMOD", "ICDE", "VLDB", "EDBT",
 21 | 				"PODS", "ICDT", "DASFAA", "SSDBM", "CIKM"};
 22 | 		String[] venues2 = {"KDD", "ICDM", "SDM", "PKDD", "PAKDD"};
 23 | 		String[] venues3 = {"IJCAI", "AAAI", "NIPS", "ICML", "ECML", 
 24 | 				"ACML", "IJCNN", "UAI", "ECAI", "COLT", "ACL", "KR"};
 25 | 		String[] venues4 = {"CVPR", "ICCV", "ECCV", "ACCV", "MM", 
 26 | 				"ICPR", "ICIP", "ICME"};
 27 | 		ArrayList<ArrayList<String>> venueLists = new ArrayList<ArrayList<String>>();
 28 | 		ArrayList<String> venueList1 = new ArrayList<String>();
 29 | 		for(int i=0;i<venues1.length;i++)
 30 | 			venueList1.add(venues1[i]);
 31 | 		venueLists.add(venueList1);
 32 | 		ArrayList<String> venueList2 = new ArrayList<String>();
 33 | 		for(int i=0;i<venues2.length;i++)
 34 | 			venueList2.add(venues2[i]);
 35 | 		venueLists.add(venueList2);
 36 | 		ArrayList<String> venueList3 = new ArrayList<String>();
 37 | 		for(int i=0;i<venues3.length;i++)
 38 | 			venueList3.add(venues3[i]);
 39 | 		venueLists.add(venueList3);
 40 | 		ArrayList<String> venueList4 = new ArrayList<String>();
 41 | 		for(int i=0;i<venues4.length;i++)
 42 | 			venueList4.add(venues4[i]);
 43 | 		venueLists.add(venueList4);
 44 | 		
 45 | 		int[] arrClass = new int[4]; // a random permutation of integers from 0 to 3 inclusive;
 46 | 		for(int i=0;i<4;i++)
 47 | 			arrClass[i] = i;
 48 | 		Random randomClass = new Random();
 49 | 		for(int i=3;i>0;i--) 
 50 | 		{
 51 | 			int randNumClass = randomClass.nextInt(i);
 52 | 			int temp = arrClass[i];
 53 | 			arrClass[i] = arrClass[randNumClass];
 54 | 			arrClass[randNumClass] = temp;
 55 | 		}
 56 | 		for(int i=0;i<4;i++)
 57 | 		{
 58 | 			for(int j=0;j<venueLists.get(arrClass[i]).size();j++)
 59 | 			{
 60 | 				String venueToken = venueLists.get(arrClass[i]).get(j);
 61 | 				if(venue.indexOf(venueToken)!=-1)
 62 | 					return arrClass[i];
 63 | 			}
 64 | 		}
 65 | 		return 0;
 66 | 	}
 67 | 
 68 | 	public static void main(String[] args) throws IOException {
 69 | 		// TODO Auto-generated method stub
 70 | 		BufferedReader brAuthor = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.author.txt"), "UTF-8"));
 71 | 		BufferedReader brVenue = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.venue.txt"), "UTF-8"));
 72 | 		BufferedReader brRef = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.ref.txt"), "UTF-8"));
 73 | 		Map<String, Integer> author2Id = new HashMap<String, Integer>();
 74 | 		Map<String, Integer> venue2Id = new HashMap<String, Integer>();
 75 | 		Map<String, Integer> paper2Id = new HashMap<String, Integer>();
 76 | 		Map<Integer, String> Id2Author = new HashMap<Integer, String>();
 77 | 		Map<Integer, String> Id2Venue = new HashMap<Integer, String>();
 78 | 		Map<Integer, String> Id2Paper = new HashMap<Integer, String>();
 79 | 		
 80 | 		String line = null;
 81 | 		int paperIndex = 0, authorIndex = 0, venueIndex = 0;
 82 | 		while((line=brAuthor.readLine())!=null)
 83 | 		{
 84 | 			int splitPos = line.indexOf("\t");
 85 | 			String paper = line.substring(0, splitPos);
 86 | 			String author = line.substring(splitPos+1);
 87 | 			if(!paper2Id.containsKey(paper))
 88 | 			{
 89 | 				paper2Id.put(paper, paperIndex);
 90 | 				Id2Paper.put(paperIndex, paper);
 91 | 				paperIndex++;
 92 | 			}
 93 | 			if(!author2Id.containsKey(author))
 94 | 			{
 95 | 				author2Id.put(author, authorIndex);
 96 | 				Id2Author.put(authorIndex, author);
 97 | 				authorIndex++;
 98 | 			}
 99 | 		}
100 | 		brAuthor.close();
101 | 		while((line=brVenue.readLine())!=null)
102 | 		{
103 | 			int splitPos = line.indexOf("\t");
104 | 			String venue = line.substring(splitPos+1);
105 | 			if(!venue2Id.containsKey(venue))
106 | 			{
107 | 				venue2Id.put(venue, venueIndex);
108 | 				Id2Venue.put(venueIndex, venue);
109 | 				venueIndex++;
110 | 			}
111 | 		}
112 | 		brVenue.close();
113 | 		int numOfPaper = paperIndex;
114 | 		int numOfAuthor = authorIndex;
115 | 		int numOfVenue = venueIndex;
116 | 
117 | 		BufferedWriter bwPaperPaper = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.paper.paper.txt"), "UTF-8"));
118 | 		while((line=brRef.readLine())!=null)
119 | 		{
120 | 			int splitPos = line.indexOf("\t");
121 | 			String paper1 = line.substring(0, splitPos);
122 | 			String paper2 = line.substring(splitPos+1);
123 | 			if(paper2Id.containsKey(paper2))
124 | 				bwPaperPaper.write(paper1+"\t"+paper2+"\n");
125 | 		}
126 | 		brRef.close();
127 | 		bwPaperPaper.close();
128 | 		
129 | 		ArrayList<ArrayList<Integer>> authorNeighbors = new ArrayList<ArrayList<Integer>>();
130 | 		ArrayList<ArrayList<Integer>> paperNeighborsVenue = new ArrayList<ArrayList<Integer>>();
131 | 		for(int i=0;i<numOfAuthor;i++)
132 | 		{
133 | 			ArrayList<Integer> neighbors = new ArrayList<Integer>();
134 | 			authorNeighbors.add(neighbors);
135 | 		}
136 | 		for(int i=0;i<numOfPaper;i++)
137 | 		{
138 | 			ArrayList<Integer> neighborsVenue = new ArrayList<Integer>();
139 | 			paperNeighborsVenue.add(neighborsVenue);
140 | 		}
141 | 		brAuthor = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.author.txt"), "UTF-8"));
142 | 		brVenue = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.venue.txt"), "UTF-8"));
143 | 		while((line=brAuthor.readLine())!=null)
144 | 		{
145 | 			int splitPos = line.indexOf("\t");
146 | 			String paper = line.substring(0, splitPos);
147 | 			String author = line.substring(splitPos+1);
148 | 			int paperId = paper2Id.get(paper);
149 | 			int authorId = author2Id.get(author);
150 | 			if(!authorNeighbors.get(authorId).contains(paperId))
151 | 				authorNeighbors.get(authorId).add(paperId);
152 | 		}
153 | 		brAuthor.close();
154 | 		while((line=brVenue.readLine())!=null)
155 | 		{
156 | 			int splitPos = line.indexOf("\t");
157 | 			String paper = line.substring(0, splitPos);
158 | 			String venue = line.substring(splitPos+1);
159 | 			int paperId = paper2Id.get(paper);
160 | 			int venueId = venue2Id.get(venue);
161 | 			if(!paperNeighborsVenue.get(paperId).contains(venueId))
162 | 				paperNeighborsVenue.get(paperId).add(venueId);
163 | 		}
164 | 		brVenue.close();
165 | 		for(int i=0;i<numOfAuthor;i++)
166 | 		{
167 | 			Collections.sort(authorNeighbors.get(i), new Comparator<Integer>() {
168 | 				public int compare(Integer a, Integer b) {
169 | 					return a.compareTo(b);
170 | 				}
171 | 			});
172 | 		}
173 | 		for(int i=0;i<numOfPaper;i++)
174 | 		{
175 | 			Collections.sort(paperNeighborsVenue.get(i), new Comparator<Integer>() {
176 | 				public int compare(Integer a, Integer b) {
177 | 					return a.compareTo(b);
178 | 				}
179 | 			});
180 | 		}
181 | 		
182 | 		int[][] venueClass = new int[numOfVenue][4];
183 | 		int[][] authorClass = new int[numOfAuthor][4];
184 | 		for(int i=0;i<numOfVenue;i++)
185 | 		{
186 | 			for(int j=0;j<4;j++)
187 | 				venueClass[i][j] = 0;
188 | 			String venue = Id2Venue.get(i);
189 | 			int label = getClass(venue);
190 | 			venueClass[i][label] = 1;
191 | 		}
192 | 		for(int i=0;i<numOfAuthor;i++)
193 | 		{
194 | 			for(int j=0;j<4;j++)
195 | 				authorClass[i][j] = 0;
196 | 			for(int j=0;j<authorNeighbors.get(i).size();j++)
197 | 			{
198 | 				int paperId = authorNeighbors.get(i).get(j);
199 | 				for(int k=0;k<paperNeighborsVenue.get(paperId).size();k++)
200 | 				{
201 | 					int venueId = paperNeighborsVenue.get(paperId).get(k);
202 | 					String venue = Id2Venue.get(venueId);
203 | 					int label = getClass(venue);
204 | 					authorClass[i][label] += 1;
205 | 				}
206 | 			}
207 | 		}
208 | 		for(int i=0;i<numOfAuthor;i++)
209 | 		{
210 | 			int[] arrClass = new int[4]; // a random permutation of integers from 0 to 3 inclusive;
211 | 			for(int j=0;j<4;j++)
212 | 				arrClass[j] = j;
213 | 			Random randomClass = new Random();
214 | 			for(int j=3;j>0;j--) 
215 | 			{
216 | 				int randNumClass = randomClass.nextInt(j);
217 | 				int temp = arrClass[j];
218 | 				arrClass[j] = arrClass[randNumClass];
219 | 				arrClass[randNumClass] = temp;
220 | 			}
221 | 			int label = -1;
222 | 			int maxNum = -1;
223 | 			for(int j=0;j<4;j++)
224 | 			{
225 | 				if(authorClass[i][arrClass[j]]>maxNum)
226 | 				{
227 | 					maxNum = authorClass[i][arrClass[j]];
228 | 					label = arrClass[j];
229 | 				}
230 | 			}
231 | 			if(label==-1)
232 | 				System.out.println("no class label for author "+i);
233 | 			else
234 | 			{
235 | 				for(int j=0;j<4;j++)
236 | 					authorClass[i][j] = 0;
237 | 				authorClass[i][label] = 1;
238 | 			}
239 | 		}
240 | 		
241 | 		BufferedWriter bwAuthorClass = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("group"+File.separator+"DBLP-V3.author.class.txt"), "UTF-8"));
242 | 		for(int i=0;i<numOfAuthor;i++)
243 | 		{
244 | 			String author = Id2Author.get(i);
245 | 			bwAuthorClass.write(author);
246 | 			for(int j=0;j<4;j++)
247 | 				bwAuthorClass.write(" "+authorClass[i][j]);
248 | 			bwAuthorClass.write("\n");
249 | 		}
250 | 		bwAuthorClass.close();
251 |   	}
252 | }
253 | 


--------------------------------------------------------------------------------
/DBLPPreProcess/src/PreProcessStep3.java:
--------------------------------------------------------------------------------
  1 | import java.io.BufferedReader;
  2 | import java.io.BufferedWriter;
  3 | import java.io.File;
  4 | import java.io.FileInputStream;
  5 | import java.io.FileOutputStream;
  6 | import java.io.IOException;
  7 | import java.io.InputStreamReader;
  8 | import java.io.OutputStreamWriter;
  9 | import java.util.ArrayList;
 10 | import java.util.Collections;
 11 | import java.util.Comparator;
 12 | import java.util.HashMap;
 13 | import java.util.Map;
 14 | import java.util.Random;
 15 | 
 16 | public class PreProcessStep3 { // generate random walk sequences, including MetaGraph guided random walk, MetaPath guided random walk, and uniform random walk
 17 | 	
 18 | 	public static Map<String, Integer> author2Id;
 19 | 	public static Map<String, Integer> venue2Id;
 20 | 	public static Map<String, Integer> paper2Id;
 21 | 	public static Map<Integer, String> Id2Author;
 22 | 	public static Map<Integer, String> Id2Venue;
 23 | 	public static Map<Integer, String> Id2Paper;
 24 | 	
 25 | 	public static ArrayList<ArrayList<Integer>> authorNeighbors;
 26 | 	public static ArrayList<ArrayList<Integer>> venueNeighbors;
 27 | 	public static ArrayList<ArrayList<Integer>> paperNeighborsAuthor;
 28 | 	public static ArrayList<ArrayList<Integer>> paperNeighborsPaper;
 29 | 	public static ArrayList<ArrayList<Integer>> paperNeighborsVenue;
 30 | 	
 31 | 	public static String metaGraphRandomWalk(int startNodeId, String type, int pathLen)
 32 | 	{
 33 | 		int curNodeId = startNodeId;
 34 | 		String curType = type;
 35 | 		int curPos = 1;
 36 | 		Random random = new Random();
 37 | 		StringBuffer path = new StringBuffer("");
 38 | 		for(int i=0;i<pathLen;i++)
 39 | 		{
 40 | 			if(curType.equals("P"))
 41 | 			{
 42 | 				String paper = Id2Paper.get(curNodeId);
 43 | 				if(i>0)
 44 | 					path.append(" ");
 45 | 				path.append(paper);
 46 | 				if(curPos==2)
 47 | 				{
 48 | 					if(paperNeighborsAuthor.get(curNodeId).size()>1&&paperNeighborsVenue.get(curNodeId).size()>0) 
 49 | 					{
 50 | 						double option = Math.random();
 51 | 						if(option<1.0/2.0)
 52 | 						{
 53 | 							int nextIndex = random.nextInt(paperNeighborsAuthor.get(curNodeId).size());
 54 | 							curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
 55 | 							curPos++;
 56 | 							curType = "A";
 57 | 						}
 58 | 						else
 59 | 						{
 60 | 							int nextIndex = random.nextInt(paperNeighborsVenue.get(curNodeId).size());
 61 | 							curNodeId = paperNeighborsVenue.get(curNodeId).get(nextIndex);
 62 | 							curPos++;
 63 | 							curType = "V";
 64 | 						}
 65 | 					}
 66 | 					else if(paperNeighborsAuthor.get(curNodeId).size()>1)
 67 | 					{
 68 | 						int nextIndex = random.nextInt(paperNeighborsAuthor.get(curNodeId).size());
 69 | 						curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
 70 | 						curPos++;
 71 | 						curType = "A";
 72 | 					}
 73 | 					else if(paperNeighborsVenue.get(curNodeId).size()>0)
 74 | 					{
 75 | 						int nextIndex = random.nextInt(paperNeighborsVenue.get(curNodeId).size());
 76 | 						curNodeId = paperNeighborsVenue.get(curNodeId).get(nextIndex);
 77 | 						curPos++;
 78 | 						curType = "V";
 79 | 					}
 80 | 					else
 81 | 						return path.toString();
 82 | 				}
 83 | 				else
 84 | 				{
 85 | 					if(paperNeighborsAuthor.get(curNodeId).size()==0)
 86 | 						return path.toString();
 87 | 					int nextIndex = random.nextInt(paperNeighborsAuthor.get(curNodeId).size());
 88 | 					curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
 89 | 					curPos++;
 90 | 					if(curPos>=5)
 91 | 						curPos = 1;
 92 | 					curType = "A";
 93 | 				}
 94 | 			}
 95 | 			else
 96 | 			{
 97 | 				if(curType.equals("A"))
 98 | 				{
 99 | 					String author = Id2Author.get(curNodeId);
100 | 					if(i>0)
101 | 						path.append(" ");
102 | 					path.append(author);
103 | 					if(authorNeighbors.get(curNodeId).size()==0)
104 | 						return path.toString();
105 | 					int nextIndex = random.nextInt(authorNeighbors.get(curNodeId).size());
106 | 					curNodeId = authorNeighbors.get(curNodeId).get(nextIndex);
107 | 					curPos++;
108 | 					curType = "P";
109 | 				}
110 | 				else if(curType.equals("V"))
111 | 				{
112 | 					String venue = Id2Venue.get(curNodeId);
113 | 					if(i>0)
114 | 						path.append(" ");
115 | 					path.append(venue);
116 | 					if(venueNeighbors.get(curNodeId).size()==0)
117 | 						return path.toString();
118 | 					int nextIndex = random.nextInt(venueNeighbors.get(curNodeId).size());
119 | 					curNodeId = venueNeighbors.get(curNodeId).get(nextIndex);
120 | 					curPos++;
121 | 					curType = "P";
122 | 				}
123 | 			}
124 | 		}
125 | 		return path.toString();
126 | 	}
127 | 	
128 | 	public static String metaPathAPVPARandomWalk(int startNodeId, String type, int pathLen)
129 | 	{
130 | 		int curNodeId = startNodeId;
131 | 		String curType = type;
132 | 		String lastType = "";
133 | 		Random random = new Random();
134 | 		StringBuffer path = new StringBuffer("");
135 | 		for(int i=0;i<pathLen;i++)
136 | 		{
137 | 			if(curType.equals("P"))
138 | 			{
139 | 				String paper = Id2Paper.get(curNodeId);
140 | 				if(i>0)
141 | 					path.append(" ");
142 | 				path.append(paper);
143 | 				if(lastType.equals("A"))
144 | 				{
145 | 					if(paperNeighborsVenue.get(curNodeId).size()==0)
146 | 						return path.toString();
147 | 					int nextIndex = random.nextInt(paperNeighborsVenue.get(curNodeId).size());
148 | 					curNodeId = paperNeighborsVenue.get(curNodeId).get(nextIndex);
149 | 					lastType = curType;
150 | 					curType = "V";
151 | 				}
152 | 				else
153 | 				{
154 | 					if(paperNeighborsAuthor.get(curNodeId).size()==0)
155 | 						return path.toString();
156 | 					int nextIndex = random.nextInt(paperNeighborsAuthor.get(curNodeId).size());
157 | 					curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
158 | 					lastType = curType;
159 | 					curType = "A";
160 | 				}
161 | 			}
162 | 			else
163 | 			{
164 | 				if(curType.equals("A"))
165 | 				{
166 | 					String author = Id2Author.get(curNodeId);
167 | 					if(i>0)
168 | 						path.append(" ");
169 | 					path.append(author);
170 | 					if(authorNeighbors.get(curNodeId).size()==0)
171 | 						return path.toString();
172 | 					int nextIndex = random.nextInt(authorNeighbors.get(curNodeId).size());
173 | 					curNodeId = authorNeighbors.get(curNodeId).get(nextIndex);
174 | 					lastType = curType;
175 | 					curType = "P";
176 | 				}
177 | 				else
178 | 				{
179 | 					String venue = Id2Venue.get(curNodeId);
180 | 					if(i>0)
181 | 						path.append(" ");
182 | 					path.append(venue);
183 | 					if(venueNeighbors.get(curNodeId).size()==0)
184 | 						return path.toString();
185 | 					int nextIndex = random.nextInt(venueNeighbors.get(curNodeId).size());
186 | 					curNodeId = venueNeighbors.get(curNodeId).get(nextIndex);
187 | 					lastType = curType;
188 | 					curType = "P";
189 | 				}
190 | 			}
191 | 		}
192 | 		return path.toString();
193 | 	}
194 | 	
195 | 	public static String metaPathAPAPARandomWalk(int startNodeId, String type, int pathLen)
196 | 	{
197 | 		int curNodeId = startNodeId;
198 | 		String curType = type;
199 | 		Random random = new Random();
200 | 		StringBuffer path = new StringBuffer("");
201 | 		for(int i=0;i<pathLen;i++)
202 | 		{
203 | 			if(curType.equals("P"))
204 | 			{
205 | 				String paper = Id2Paper.get(curNodeId);
206 | 				if(i>0)
207 | 					path.append(" ");
208 | 				path.append(paper);
209 | 				if(paperNeighborsAuthor.get(curNodeId).size()<=1)
210 | 					return path.toString();
211 | 				int nextIndex = random.nextInt(paperNeighborsAuthor.get(curNodeId).size());
212 | 				curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
213 | 				curType = "A";
214 | 			}
215 | 			else
216 | 			{
217 | 				String author = Id2Author.get(curNodeId);
218 | 				if(i>0)
219 | 					path.append(" ");
220 | 				path.append(author);
221 | 				if(authorNeighbors.get(curNodeId).size()==0)
222 | 					return path.toString();
223 | 				int nextIndex = random.nextInt(authorNeighbors.get(curNodeId).size());
224 | 				curNodeId = authorNeighbors.get(curNodeId).get(nextIndex);
225 | 				curType = "P";
226 | 			}
227 | 		}
228 | 		return path.toString();
229 | 	}
230 | 	
231 | 	public static String uniformRandomWalk(int startNodeId, String type, int pathLen)
232 | 	{
233 | 		int curNodeId = startNodeId;
234 | 		String curType = type;
235 | 		Random random = new Random();
236 | 		StringBuffer path = new StringBuffer("");
237 | 		for(int i=0;i<pathLen;i++)
238 | 		{
239 | 			if(curType.equals("P"))
240 | 			{
241 | 				String paper = Id2Paper.get(curNodeId);
242 | 				if(i>0)
243 | 					path.append(" ");
244 | 				path.append(paper);
245 | 				int authorSize = paperNeighborsAuthor.get(curNodeId).size();
246 | 				int paperSize = paperNeighborsPaper.get(curNodeId).size();
247 | 				int venueSize = paperNeighborsVenue.get(curNodeId).size();
248 | 				if(authorSize+paperSize+venueSize==0)
249 | 					return path.toString();
250 | 				int nextIndex = random.nextInt(authorSize+paperSize+venueSize);
251 | 				if(nextIndex<authorSize)
252 | 				{
253 | 					curNodeId = paperNeighborsAuthor.get(curNodeId).get(nextIndex);
254 | 					curType = "A";
255 | 				}
256 | 				else if(nextIndex<authorSize+paperSize)
257 | 				{
258 | 					curNodeId = paperNeighborsPaper.get(curNodeId).get(nextIndex-authorSize);
259 | 					curType = "P";
260 | 				}
261 | 				else if(nextIndex<authorSize+paperSize+venueSize)
262 | 				{
263 | 					curNodeId = paperNeighborsVenue.get(curNodeId).get(nextIndex-authorSize-paperSize);
264 | 					curType = "V";
265 | 				}
266 | 			}
267 | 			else
268 | 			{
269 | 				if(curType.equals("A"))
270 | 				{
271 | 					String author = Id2Author.get(curNodeId);
272 | 					if(i>0)
273 | 						path.append(" ");
274 | 					path.append(author);
275 | 					if(authorNeighbors.get(curNodeId).size()==0)
276 | 						return path.toString();
277 | 					int nextIndex = random.nextInt(authorNeighbors.get(curNodeId).size());
278 | 					curNodeId = authorNeighbors.get(curNodeId).get(nextIndex);
279 | 					curType = "P";
280 | 				}
281 | 				else if(curType.equals("V"))
282 | 				{
283 | 					String venue = Id2Venue.get(curNodeId);
284 | 					if(i>0)
285 | 						path.append(" ");
286 | 					path.append(venue);
287 | 					if(venueNeighbors.get(curNodeId).size()==0)
288 | 						return path.toString();
289 | 					int nextIndex = random.nextInt(venueNeighbors.get(curNodeId).size());
290 | 					curNodeId = venueNeighbors.get(curNodeId).get(nextIndex);
291 | 					curType = "P";
292 | 				}
293 | 			}
294 | 		}
295 | 		return path.toString();
296 | 	}
297 | 	
298 | 	public static void main(String[] args) throws IOException {
299 | 		// TODO Auto-generated method stub
300 | 		BufferedReader brPaperAuthor = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.author.txt"), "UTF-8"));
301 | 		BufferedReader brPaperVenue = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.venue.txt"), "UTF-8"));
302 | 		BufferedReader brPaperPaper = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.paper.txt"), "UTF-8"));
303 | 		
304 | 		author2Id = new HashMap<String, Integer>();
305 | 		venue2Id = new HashMap<String, Integer>();
306 | 		paper2Id = new HashMap<String, Integer>();
307 | 		Id2Author = new HashMap<Integer, String>();
308 | 		Id2Venue = new HashMap<Integer, String>();
309 | 		Id2Paper = new HashMap<Integer, String>();
310 | 		
311 | 		String line = null;
312 | 		int paperIndex = 0, authorIndex = 0, venueIndex = 0;
313 | 		while((line=brPaperAuthor.readLine())!=null)
314 | 		{
315 | 			int splitPos = line.indexOf("\t");
316 | 			String paper = line.substring(0, splitPos);
317 | 			String author = line.substring(splitPos+1);
318 | 			if(!paper2Id.containsKey(paper))
319 | 			{
320 | 				paper2Id.put(paper, paperIndex);
321 | 				Id2Paper.put(paperIndex, paper);
322 | 				paperIndex++;
323 | 			}
324 | 			if(!author2Id.containsKey(author))
325 | 			{
326 | 				author2Id.put(author, authorIndex);
327 | 				Id2Author.put(authorIndex, author);
328 | 				authorIndex++;
329 | 			}
330 | 		}
331 | 		brPaperAuthor.close();
332 | 		while((line=brPaperVenue.readLine())!=null)
333 | 		{
334 | 			int splitPos = line.indexOf("\t");
335 | 			String venue = line.substring(splitPos+1);
336 | 			if(!venue2Id.containsKey(venue))
337 | 			{
338 | 				venue2Id.put(venue, venueIndex);
339 | 				Id2Venue.put(venueIndex, venue);
340 | 				venueIndex++;
341 | 			}
342 | 		}
343 | 		brPaperVenue.close();
344 | 		int numOfPaper = paperIndex;
345 | 		int numOfAuthor = authorIndex;
346 | 		int numOfVenue = venueIndex;
347 | 		
348 | 		authorNeighbors = new ArrayList<ArrayList<Integer>>();
349 | 		venueNeighbors = new ArrayList<ArrayList<Integer>>();
350 | 		paperNeighborsAuthor = new ArrayList<ArrayList<Integer>>();
351 | 		paperNeighborsPaper = new ArrayList<ArrayList<Integer>>();
352 | 		paperNeighborsVenue = new ArrayList<ArrayList<Integer>>();
353 | 		
354 | 		for(int i=0;i<numOfAuthor;i++)
355 | 		{
356 | 			ArrayList<Integer> neighbors = new ArrayList<Integer>();
357 | 			authorNeighbors.add(neighbors);
358 | 		}
359 | 		for(int i=0;i<numOfVenue;i++)
360 | 		{
361 | 			ArrayList<Integer> neighbors = new ArrayList<Integer>();
362 | 			venueNeighbors.add(neighbors);
363 | 		}
364 | 		for(int i=0;i<numOfPaper;i++)
365 | 		{
366 | 			ArrayList<Integer> neighborsAuthor = new ArrayList<Integer>();
367 | 			paperNeighborsAuthor.add(neighborsAuthor);
368 | 			ArrayList<Integer> neighborsPaper = new ArrayList<Integer>();
369 | 			paperNeighborsPaper.add(neighborsPaper);
370 | 			ArrayList<Integer> neighborsVenue = new ArrayList<Integer>();
371 | 			paperNeighborsVenue.add(neighborsVenue);
372 | 		}
373 | 
374 | 		brPaperAuthor = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.author.txt"), "UTF-8"));
375 | 		brPaperVenue = new BufferedReader(new InputStreamReader(new FileInputStream("dataset"+File.separator+"DBLP-V3.paper.venue.txt"), "UTF-8"));
376 | 		while((line=brPaperAuthor.readLine())!=null)
377 | 		{
378 | 			int splitPos = line.indexOf("\t");
379 | 			String paper = line.substring(0, splitPos);
380 | 			String author = line.substring(splitPos+1);
381 | 			int paperId = paper2Id.get(paper);
382 | 			int authorId = author2Id.get(author);
383 | 			if(!authorNeighbors.get(authorId).contains(paperId))
384 | 				authorNeighbors.get(authorId).add(paperId);
385 | 			if(!paperNeighborsAuthor.get(paperId).contains(authorId))
386 | 				paperNeighborsAuthor.get(paperId).add(authorId);
387 | 		}
388 | 		brPaperAuthor.close();
389 | 		while((line=brPaperPaper.readLine())!=null)
390 | 		{
391 | 			int splitPos = line.indexOf("\t");
392 | 			String paper1 = line.substring(0, splitPos);
393 | 			String paper2 = line.substring(splitPos+1);
394 | 			int paperId1 = paper2Id.get(paper1);
395 | 			int paperId2 = paper2Id.get(paper2);
396 | 			if(!paperNeighborsPaper.get(paperId1).contains(paperId2))
397 | 				paperNeighborsPaper.get(paperId1).add(paperId2);
398 | 			if(!paperNeighborsPaper.get(paperId2).contains(paperId1))
399 | 				paperNeighborsPaper.get(paperId2).add(paperId1);
400 | 		}
401 | 		brPaperPaper.close();
402 | 		while((line=brPaperVenue.readLine())!=null)
403 | 		{
404 | 			int splitPos = line.indexOf("\t");
405 | 			String paper = line.substring(0, splitPos);
406 | 			String venue = line.substring(splitPos+1);
407 | 			int paperId = paper2Id.get(paper);
408 | 			int venueId = venue2Id.get(venue);
409 | 			if(!venueNeighbors.get(venueId).contains(paperId))
410 | 				venueNeighbors.get(venueId).add(paperId);
411 | 			if(!paperNeighborsVenue.get(paperId).contains(venueId))
412 | 				paperNeighborsVenue.get(paperId).add(venueId);
413 | 		}
414 | 		brPaperVenue.close();
415 | 		
416 | 		int[] arrPaper1 = new int[numOfPaper];
417 | 		for(int i=0;i<numOfPaper;i++)
418 | 			arrPaper1[i] = i;
419 | 		Random randomPaper1 = new Random();
420 | 		for(int i=numOfPaper-1;i>0;i--) {
421 | 			int randNumPaper1 = randomPaper1.nextInt(i);
422 | 			int tempPaper1 = arrPaper1[i];
423 | 			arrPaper1[i] = arrPaper1[randNumPaper1];
424 | 			arrPaper1[randNumPaper1] = tempPaper1;
425 | 		}
426 | 		
427 | 		for(int i=0;i<numOfPaper;i++) {
428 | 			if(i%5==0) {
429 | 				int paperId = arrPaper1[i];
430 | 				for(int j=0;j<paperNeighborsVenue.get(paperId).size();j++) {
431 | 					int venueId = paperNeighborsVenue.get(paperId).get(j);
432 | 					int paperPos = -1;
433 | 					for(int k=0;k<venueNeighbors.get(venueId).size();k++) {
434 | 						if(venueNeighbors.get(venueId).get(k)==paperId) {
435 | 							paperPos = k;
436 | 							break;
437 | 						}
438 | 					}
439 | 					if(paperPos!=-1)
440 | 						venueNeighbors.get(venueId).remove(paperPos);
441 | 				}
442 | 				paperNeighborsVenue.get(paperId).clear();
443 | 			}
444 | 		}
445 | 		
446 | 		for(int i=0;i<numOfAuthor;i++)
447 | 		{
448 | 			Collections.sort(authorNeighbors.get(i), new Comparator<Integer>() {
449 | 				public int compare(Integer a, Integer b) {
450 | 					return a.compareTo(b);
451 | 				}
452 | 			});
453 | 		}
454 | 		for(int i=0;i<numOfVenue;i++)
455 | 		{
456 | 			Collections.sort(venueNeighbors.get(i), new Comparator<Integer>() {
457 | 				public int compare(Integer a, Integer b) {
458 | 					return a.compareTo(b);
459 | 				}
460 | 			});
461 | 		}
462 | 		
463 | 		for(int i=0;i<numOfPaper;i++)
464 | 		{
465 | 			Collections.sort(paperNeighborsAuthor.get(i), new Comparator<Integer>() {
466 | 				public int compare(Integer a, Integer b) {
467 | 					return a.compareTo(b);
468 | 				}
469 | 			});
470 | 			Collections.sort(paperNeighborsPaper.get(i), new Comparator<Integer>() {
471 | 				public int compare(Integer a, Integer b) {
472 | 					return a.compareTo(b);
473 | 				}
474 | 			});
475 | 			Collections.sort(paperNeighborsVenue.get(i), new Comparator<Integer>() {
476 | 				public int compare(Integer a, Integer b) {
477 | 					return a.compareTo(b);
478 | 				}
479 | 			});
480 | 		}
481 | 		
482 | 		BufferedWriter bwMetaGraphRandWalk = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("randomwalk"+File.separator+"DBLP-V3.meta.graph.random.walk.txt"), "UTF-8"));
483 | 		BufferedWriter bwMetaPathRandWalkAPVPA = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("randomwalk"+File.separator+"DBLP-V3.meta.path.random.walk.apvpa.txt"), "UTF-8"));
484 | 		BufferedWriter bwMetaPathRandWalkAPAPA = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("randomwalk"+File.separator+"DBLP-V3.meta.path.random.walk.apapa.txt"), "UTF-8"));
485 | 		BufferedWriter bwMetaPathRandWalkMix = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("randomwalk"+File.separator+"DBLP-V3.meta.path.random.walk.mix.txt"), "UTF-8"));
486 | 		BufferedWriter bwUniformRandWalk = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("randomwalk"+File.separator+"DBLP-V3.uniform.random.walk.txt"), "UTF-8"));
487 | 		int pathLen = 100;
488 | 		int numOfWalk = 80;
489 | 		for(int i=0;i<numOfWalk;i++)
490 | 		{
491 | 			System.out.println("rand walking iteration = "+i);
492 | 			int[] arr = new int[numOfAuthor]; // a random permutation of integers from 0 to numOfAuthor-1 inclusive;
493 | 			for(int j=0;j<numOfAuthor;j++)
494 | 				arr[j] = j;
495 | 			Random random = new Random();
496 | 			for(int j=numOfAuthor-1;j>0;j--) {
497 | 				int randNum = random.nextInt(j);
498 | 				int temp = arr[j];
499 | 				arr[j] = arr[randNum];
500 | 				arr[randNum] = temp;
501 | 			}
502 | 			for(int j=0;j<numOfAuthor;j++)
503 | 			{
504 | 				String path = null;
505 | 				String type = "A";
506 | 				int startNodeId = arr[j];
507 | 				path = metaGraphRandomWalk(startNodeId, type, pathLen);
508 | 				bwMetaGraphRandWalk.write(path+"\n");
509 | 				path = metaPathAPVPARandomWalk(startNodeId, type, pathLen);
510 | 				bwMetaPathRandWalkAPVPA.write(path+"\n");
511 | 				path = metaPathAPAPARandomWalk(startNodeId, type, pathLen);
512 | 				bwMetaPathRandWalkAPAPA.write(path+"\n");
513 | 				path = uniformRandomWalk(startNodeId, type, pathLen);
514 | 				bwUniformRandWalk.write(path+"\n");
515 | 			}
516 | 		}
517 | 		int[] arr1 = new int[numOfWalk]; // a random permutation of integers from 0 to numOfWalk-1 inclusive;
518 | 		for(int i=0;i<numOfWalk;i++)
519 | 			arr1[i] = i;
520 | 		Random random1 = new Random();
521 | 		for(int i=numOfWalk-1;i>0;i--) {
522 | 			int randNum1 = random1.nextInt(i);
523 | 			int temp1 = arr1[i];
524 | 			arr1[i] = arr1[randNum1];
525 | 			arr1[randNum1] = temp1;
526 | 		}
527 | 		for(int i=0;i<numOfWalk;i++)
528 | 		{
529 | 			System.out.println("rand walking iteration = "+i);
530 | 			int[] arr2 = new int[numOfAuthor]; // a random permutation of integers from 1 to numOfAuthor inclusive;
531 | 			for(int j=0;j<numOfAuthor;j++)
532 | 				arr2[j] = j;
533 | 			Random random2 = new Random();
534 | 			for(int j=numOfAuthor-1;j>0;j--) {
535 | 				int randNum2 = random2.nextInt(j);
536 | 				int temp2 = arr2[j];
537 | 				arr2[j] = arr2[randNum2];
538 | 				arr2[randNum2] = temp2;
539 | 			}
540 | 			for(int j=0;j<numOfAuthor;j++)
541 | 			{
542 | 				String path = null;
543 | 				String type = "A";
544 | 				int startNodeId = arr2[j];
545 | 				if(arr1[i]%2==0)
546 | 				{
547 | 					path = metaPathAPVPARandomWalk(startNodeId, type, pathLen);
548 | 					bwMetaPathRandWalkMix.write(path+"\n");
549 | 				}
550 | 				else
551 | 				{
552 | 					path = metaPathAPAPARandomWalk(startNodeId, type, pathLen);
553 | 					bwMetaPathRandWalkMix.write(path+"\n");
554 | 				}
555 | 			}
556 | 		}
557 | 		bwMetaGraphRandWalk.close();
558 | 		bwMetaPathRandWalkAPVPA.close();
559 | 		bwMetaPathRandWalkAPAPA.close();
560 | 		bwUniformRandWalk.close();
561 | 		bwMetaPathRandWalkMix.close();
562 | 		
563 | 		BufferedWriter bwNetwork = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.graph.txt"), "UTF-8"));
564 | 		for(int i=0;i<numOfPaper;i++)
565 | 		{
566 | 			int curNodeId = i;
567 | 			String curPaper = Id2Paper.get(curNodeId);
568 | 			for(int j=0;j<paperNeighborsPaper.get(curNodeId).size();j++)
569 | 			{
570 | 				int neighNodeId = paperNeighborsPaper.get(curNodeId).get(j);
571 | 				String neighPaper = Id2Paper.get(neighNodeId);
572 | 				bwNetwork.write(curPaper+" "+neighPaper+" 1\n");
573 | 			}
574 | 			for(int j=0;j<paperNeighborsAuthor.get(curNodeId).size();j++)
575 | 			{
576 | 				int neighNodeId = paperNeighborsAuthor.get(curNodeId).get(j);
577 | 				String neighAuthor = Id2Author.get(neighNodeId);
578 | 				bwNetwork.write(curPaper+" "+neighAuthor+" 1\n");
579 | 			}
580 | 			for(int j=0;j<paperNeighborsVenue.get(curNodeId).size();j++)
581 | 			{
582 | 				int neighNodeId = paperNeighborsVenue.get(curNodeId).get(j);
583 | 				String neighVenue = Id2Venue.get(neighNodeId);
584 | 				bwNetwork.write(curPaper+" "+neighVenue+" 1\n");
585 | 			}
586 | 		}
587 | 		for(int i=0;i<numOfAuthor;i++)
588 | 		{
589 | 			int curNodeId = i;
590 | 			String curAuthor = Id2Author.get(curNodeId);
591 | 			for(int j=0;j<authorNeighbors.get(curNodeId).size();j++)
592 | 			{
593 | 				int neighNodeId = authorNeighbors.get(curNodeId).get(j);
594 | 				String neighPaper = Id2Paper.get(neighNodeId);
595 | 				bwNetwork.write(curAuthor+" "+neighPaper+" 1\n");
596 | 			}
597 | 		}
598 | 		for(int i=0;i<numOfVenue;i++)
599 | 		{
600 | 			int curNodeId = i;
601 | 			String curVeune = Id2Venue.get(curNodeId);
602 | 			for(int j=0;j<venueNeighbors.get(curNodeId).size();j++)
603 | 			{
604 | 				int neighNodeId = venueNeighbors.get(curNodeId).get(j);
605 | 				String neighPaper = Id2Paper.get(neighNodeId);
606 | 				bwNetwork.write(curVeune+" "+neighPaper+" 1\n");
607 | 			}
608 | 		}
609 | 		bwNetwork.close();
610 | 		
611 | 		BufferedWriter bwLog = new BufferedWriter(new OutputStreamWriter(new FileOutputStream("dataset"+File.separator+"DBLP-V3.log.txt"), "UTF-8"));
612 | 		bwLog.write("numOfAuthor = "+numOfAuthor+"\n");
613 | 		bwLog.write("numOfPaper = "+numOfPaper+"\n");
614 | 		bwLog.write("numOfVenue = "+numOfVenue+"\n");
615 | 		int numOfPaperAuthorRelation = 0;
616 | 		int numOfPaperVenueRelation = 0;
617 | 		int numOfPaperPaperRelation = 0;
618 | 		for(int i=0;i<numOfPaper;i++)
619 | 		{
620 | 			numOfPaperAuthorRelation += paperNeighborsAuthor.get(i).size();
621 | 			numOfPaperPaperRelation += paperNeighborsPaper.get(i).size();
622 | 			numOfPaperVenueRelation += paperNeighborsVenue.get(i).size();
623 | 		}
624 | 		bwLog.write("numOfPaperAuthorRelation = "+numOfPaperAuthorRelation+"\n");
625 | 		bwLog.write("numOfPaperPaperRelation = "+numOfPaperPaperRelation+"\n");
626 | 		bwLog.write("numOfPaperVenueRelation = "+numOfPaperVenueRelation+"\n");
627 | 		bwLog.close();
628 |   	}
629 | }
630 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # MetaGraph2Vec
 2 | 
 3 | Code for PAKDD-2018 paper "MetaGraph2Vec: Complex Semantic Path Augmented Heterogeneous Network Embedding"
 4 | 
 5 | Authors: Daokun Zhang, Jie Yin, Xingquan Zhu and Chengqi Zhang
 6 | 
 7 | Contact: Daokun Zhang (daokunzhang2015@gmail.com)
 8 | 
 9 | This project contains two folders:
10 | 
11 | 1) The folder "DBLPPreProcess" is the Java project for data preprocessing. It contains three files
12 | 
13 | 	"PreProcessStep1", used to extract subgraph form the "DBLP-Citation-Network-V3.txt" network;
14 | 
15 | 	"PreProcessStep2", used to obtain the class labels for authors;
16 | 
17 | 	"PreProcessStep3", used to generate random walks from the extracted subgraph, including the metagraph guided random walks, metapath guided random walks and uniform random walks that ignore the heterogeneity of nodes.
18 | 
19 | To run this project, firstly download the "DBLP-Citation-Network-V3.txt" file from https://aminer.org/citation, and move it to the "DBLPPreProcess/dataset" folder, then run "PreProcessStep1", "PreProcessStep2" and "PreProcessStep3" sequentially. The class label file will be output to the folder "DBLPPreProcess/group", and the random walk files will be output to the folder "DBLPPreProcess/randomwalk".
20 | 
21 | 2) The folder "RandomWalk2VecRun" contains the program for learning node embeddings from the generated random walk sequences. Please run "compile.sh" to compile the "randomwWalk2vec.c" file to its corresponding executable file. To run this program, move the generated random walk files from "DBLPPreProcess/randomwalk" to this folder and run the "RandomWalk2VecRun.sh" file.
22 | 
23 | In this folder, the "randomWalk2vec" program is used to learn node embeddings from random walk sequences with Heterogeneous Skip-Gram or Homogeneous Skip-Gram.
24 | 
25 | The options of randomWalk2Vec are as follows:
26 | 
27 | 	-train <file>
28 | 		Use random walk sequences from <file> to train the model
29 | 	-output <file>
30 | 		Use <file> to save the learned node vector-format representations
31 | 	-size <int>
32 | 		Set the dimension of learned node representations; default is 128
33 | 	-window <int>
34 | 		Set the window size for collecting node context pairs; default is 5
35 | 	-pp <int>
36 | 		Use Heterogeneous Skip-Gram model (the ++ version) or not; default is 1 (++ version); otherwise, use 0 (Homogeneous Skip-Gram)
37 | 	-prefixes <string>
38 | 		Prefixes of node Ids for specifying node types, e.g., ap with a for author and p for paper
39 | 	-objtype
40 | 		The index of the objective node type in the prefixes list, for which representations to be learned
41 | 	-alpha <float>
42 | 		Set the starting learning rate; default is 0.025
43 | 	-negative <int>
44 | 		Number of negative examples; default is 5, common values are 3 - 10
45 | 	-samples <int>
46 | 		Set the number of iterations for stochastic gradient descent as <int> Million; default is 100
47 | 
48 | If you find this project is useful, please cite this paper:
49 | 
50 | 	@inproceedings{zhang2018metagraph2vec,
51 | 		title={MetaGraph2Vec: Complex Semantic Path Augmented Heterogeneous Network Embedding},
52 | 		author={Zhang, Daokun and Yin, Jie and Zhu, Xingquan and Zhang, Chengqi},
53 | 		booktitle={Pacific-Asia Conference on Knowledge Discovery and Data Mining},
54 | 		pages={196--208},
55 | 		year={2018},
56 | 		organization={Springer}
57 | 	}
58 | # Note
59 | The randomWalk2vec.c is compiled on a Linux system with RAND_MAX taking value 2147483647. If you compile randomWalk2vec.c on your system, please carefully check the value of RAND_MAX to make sure it is large enough for the correctness of alias
60 | table sampling.
61 | 


--------------------------------------------------------------------------------
/RandomWalk2Vec/RandomWalk2VecRun.sh:
--------------------------------------------------------------------------------
 1 | #!/bin/bash
 2 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.graph.random.walk.txt -output DBLP-V3.meta.graph.pp1.emb.txt -pp 1 -prefixes apv -objtype 0
 3 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.graph.random.walk.txt -output DBLP-V3.meta.graph.pp0.emb.txt -pp 0 -prefixes apv -objtype 0
 4 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.apapa.txt -output DBLP-V3.meta.path.apapa.pp1.emb.txt -pp 1 -prefixes apv -objtype 0 
 5 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.apapa.txt -output DBLP-V3.meta.path.apapa.pp0.emb.txt -pp 0 -prefixes apv -objtype 0
 6 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.apvpa.txt -output DBLP-V3.meta.path.apvpa.pp1.emb.txt -pp 1 -prefixes apv -objtype 0
 7 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.apvpa.txt -output DBLP-V3.meta.path.apvpa.pp0.emb.txt -pp 0 -prefixes apv -objtype 0
 8 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.mix.txt -output DBLP-V3.meta.path.mix.pp1.emb.txt -pp 1 -prefixes apv -objtype 0
 9 | ./randomWalk2vec -size 128 -train DBLP-V3.meta.path.random.walk.mix.txt -output DBLP-V3.meta.path.mix.pp0.emb.txt -pp 0 -prefixes apv -objtype 0
10 | ./randomWalk2vec -size 128 -train DBLP-V3.uniform.random.walk.txt -output DBLP-V3.uniform.pp1.emb.txt -pp 1 -prefixes apv -objtype 0
11 | ./randomWalk2vec -size 128 -train DBLP-V3.uniform.random.walk.txt -output DBLP-V3.uniform.pp0.emb.txt -pp 0 -prefixes apv -objtype 0
12 | 


--------------------------------------------------------------------------------
/RandomWalk2Vec/compile.sh:
--------------------------------------------------------------------------------
1 | #!/bin/bash
2 | gcc -Wall -g  -c "randomWalk2vec.c" -o randomWalk2vec.o
3 | g++  -o randomWalk2vec randomWalk2vec.o
4 | 
5 | 


--------------------------------------------------------------------------------
/RandomWalk2Vec/randomWalk2vec.c:
--------------------------------------------------------------------------------
  1 | //  The randomWalk2vec project was built upon the The metapath2vec.cpp code from https://ericdongyx.github.io/metapath2vec/m2v.html
  2 | 
  3 | //  Modifications Copyright (C) 2018 <daokunzhang2015@gmail.com>
  4 | //
  5 | //  Licensed under the Apache License, Version 2.0 (the "License");
  6 | //  you may not use this file except in compliance with the License.
  7 | //  You may obtain a copy of the License at
  8 | //
  9 | //      http://www.apache.org/licenses/LICENSE-2.0
 10 | //
 11 | //  Unless required by applicable law or agreed to in writing, software
 12 | //  distributed under the License is distributed on an "AS IS" BASIS,
 13 | //  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 14 | //  See the License for the specific language governing permissions and
 15 | //  limitations under the License.
 16 | 
 17 | //  The metapath2vec.cpp code was built upon the word2vec.c from https://code.google.com/archive/p/word2vec/
 18 | 
 19 | //  Modifications Copyright (C) 2016 <ericdongyx@gmail.com>
 20 | //
 21 | //  Licensed under the Apache License, Version 2.0 (the "License");
 22 | //  you may not use this file except in compliance with the License.
 23 | //  You may obtain a copy of the License at
 24 | //
 25 | //      http://www.apache.org/licenses/LICENSE-2.0
 26 | //
 27 | //  Unless required by applicable law or agreed to in writing, software
 28 | //  distributed under the License is distributed on an "AS IS" BASIS,
 29 | //  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 30 | //  See the License for the specific language governing permissions and
 31 | //  limitations under the License.
 32 | 
 33 | //  Copyright 2013 Google Inc. All Rights Reserved.
 34 | //
 35 | //  Licensed under the Apache License, Version 2.0 (the "License");
 36 | //  you may not use this file except in compliance with the License.
 37 | //  You may obtain a copy of the License at
 38 | //
 39 | //      http://www.apache.org/licenses/LICENSE-2.0
 40 | //
 41 | //  Unless required by applicable law or agreed to in writing, software
 42 | //  distributed under the License is distributed on an "AS IS" BASIS,
 43 | //  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 44 | //  See the License for the specific language governing permissions and
 45 | //  limitations under the License.
 46 | 
 47 | #include <stdio.h>
 48 | #include <stdlib.h>
 49 | #include <string.h>
 50 | #include <math.h>
 51 | #include <time.h>
 52 | 
 53 | #define MAX_STRING 100
 54 | #define EXP_TABLE_SIZE 1000
 55 | #define MAX_EXP 6
 56 | #define MAX_WALK_LENGTH 1000
 57 | 
 58 | const long long node_hash_size = 10000000;  // Maximum 10M nodes
 59 | const long long node_context_hash_size = 1000000000; // Maximum 1G node context pairs
 60 | 
 61 | struct Node
 62 | {
 63 |     long long cn;
 64 |     char *node_str;
 65 | };
 66 | 
 67 | struct Node_Context
 68 | {
 69 |     long long cn;
 70 |     long long source, target;
 71 | };
 72 | 
 73 | char train_file[MAX_STRING], output_file[MAX_STRING];
 74 | struct Node *node_list;
 75 | long long window = 5;
 76 | long long pp = 1; // pp = 1 with Heterogeneous Skip-Gram, pp = 0 with Homogeneous Skip-Gram
 77 | long long *node_hash;
 78 | long long node_list_max_size = 100000, node_list_size = 0, layer1_size = 128;
 79 | long long node_occur_num = 0;
 80 | double alpha = 0.025, starting_alpha;
 81 | double *syn0, *syn1, *syn1neg, *expTable;
 82 | double **type_syn1;
 83 | clock_t start, finish;
 84 | 
 85 | long long type_num;
 86 | long long **type_tables, *type_counts, *type_indices;
 87 | char prefixes[MAX_STRING];
 88 | long long obj_type;
 89 | 
 90 | long long **type_node_hash;
 91 | long long *type_node_list_size, *type_node_index;
 92 | struct Node **type_node_list;
 93 | 
 94 | long long negative = 5;
 95 | const long table_size = 1e8;
 96 | long long *table;
 97 | 
 98 | long long *node_context_hash;
 99 | long long node_context_list_size, node_context_list_max_size;
100 | struct Node_Context *node_context_list;
101 | 
102 | // Parameters for node context pair sampling
103 | long long *alias;
104 | double *prob;
105 | 
106 | long long total_samples = 100;
107 | 
108 | long long GetTypeId(char prefix)
109 | {
110 |     long long i;
111 |     for (i = 0; i < type_num; i++)
112 |     {
113 |         if (prefix == prefixes[i])
114 |             break;
115 |     }
116 |     if (i >= type_num)
117 |         return -1;
118 |     else
119 |         return i;
120 | }
121 | 
122 | void NodeTypeNeg()
123 | {
124 |     long long i;
125 |     long long target, type_Id;
126 |     type_tables = (long long **)malloc(type_num * sizeof(long long *));
127 |     type_counts = (long long *)malloc(type_num * sizeof(long long));
128 |     type_indices = (long long *)malloc(type_num * sizeof(long long));
129 |     if (type_tables == NULL || type_counts == NULL || type_indices == NULL)
130 |     {
131 |         printf("Memory allocation failed\n");
132 |         exit(1);
133 |     }
134 |     for (i = 0; i < type_num; i++)
135 |     {
136 |         type_counts[i] = 0;
137 |         type_indices[i] = 0;
138 |     }
139 |     for (i = 0; i < table_size; i++)
140 |     {
141 |         target = table[i];
142 |         type_Id = GetTypeId(node_list[target].node_str[0]);
143 |         if (type_Id == -1)
144 |         {
145 |             printf("Unrecognised type\n");
146 |             exit(1);
147 |         }
148 |         type_counts[type_Id]++;
149 |     }
150 |     for (i = 0; i< type_num; i++)
151 |     {
152 |         type_tables[i] = (long long *)malloc(type_counts[i] * sizeof(long long));
153 |         if (type_tables[i] == NULL)
154 |         {
155 |             printf("Memory allocation failed\n");
156 |             exit(1);
157 |         }
158 |     }
159 |     for (i = 0; i < table_size; i++)
160 |     {
161 |         target = table[i];
162 |         type_Id = GetTypeId(node_list[target].node_str[0]);
163 |         type_tables[type_Id][type_indices[type_Id]] = target;
164 |         type_indices[type_Id]++;
165 |     }
166 |     for (i = 0; i < type_num; i++)
167 |         printf("type %c table size: %lld\n", prefixes[i], type_counts[i]);
168 | }
169 | 
170 | void InitUnigramTable()
171 | {
172 |     long long a, i;
173 |     double train_nodes_pow = 0.0, d1, power = 0.75;
174 |     table = (long long *)malloc(table_size * sizeof(long long));
175 |     if (table == NULL)
176 |     {
177 |         printf("Memory allocation failed\n");
178 |         exit(1);
179 |     }
180 |     for (a = 0; a < node_list_size; a++) train_nodes_pow += pow(node_list[a].cn, power);
181 |     i = 0;
182 |     d1 = pow(node_list[i].cn, power) / train_nodes_pow;
183 |     for (a = 0; a < table_size; a++)
184 |     {
185 |         table[a] = i;
186 |         if (a / (double)table_size > d1)
187 |         {
188 |             i++;
189 |             d1 += pow(node_list[i].cn, power) / (double)train_nodes_pow;
190 |         }
191 |         if (i >= node_list_size) i = node_list_size - 1;
192 |     }
193 |     if (pp == 1)
194 |         NodeTypeNeg();
195 | }
196 | 
197 | // Reads a single node from a file, assuming space + tab + EOL to be node name boundaries
198 | long long ReadNode(char *node_str, FILE *fin) // return 1, if the boundary is '\n', otherwise 0
199 | {
200 |     long long a = 0, ch, flag = 0;
201 |     while (!feof(fin))
202 |     {
203 |         ch = fgetc(fin);
204 |         if (ch == 13) continue;
205 |         if ((ch == ' ') || (ch == '\t') || (ch == '\n'))
206 |         {
207 |             if (a > 0)
208 |             {
209 |                 if (ch == '\n') flag = 1;
210 |                 break;
211 |             }
212 |             continue;
213 |         }
214 |         node_str[a] = ch;
215 |         a++;
216 |         if (a >= MAX_STRING - 1) a--;
217 |     }
218 |     node_str[a] = 0;
219 |     return flag;
220 | }
221 | 
222 | // Return hash value of a node
223 | long long GetNodeHash(char *node_str)
224 | {
225 |     long long a, hash = 0;
226 |     for (a = 0; a < strlen(node_str); a++)
227 |     {
228 |         hash = hash * 257 + node_str[a];
229 |         hash = hash % node_hash_size;
230 |     }
231 |     return hash;
232 | }
233 | 
234 | // Return hash value for a node context pair
235 | long long GetNodeContextHash(char *node_str, char *context_node_str)
236 | {
237 |     long long a, hash = 0;
238 |     for (a = 0; a < strlen(node_str); a++)
239 |     {
240 |         hash = hash * 257 + node_str[a];
241 |         hash = hash % node_context_hash_size;
242 |     }
243 |     for (a = 0; a < strlen(context_node_str); a++)
244 |     {
245 |         hash = hash * 257 + context_node_str[a];
246 |         hash = hash % node_context_hash_size;
247 |     }
248 |     return hash;
249 | }
250 | 
251 | // Return position of a node in the node list; if the node is not found, returns -1
252 | long long SearchNode(char *node_str)
253 | {
254 |     long long hash = GetNodeHash(node_str);
255 |     while (1)
256 |     {
257 |         if (node_hash[hash] == -1) return -1;
258 |         if (!strcmp(node_str, node_list[node_hash[hash]].node_str)) return node_hash[hash];
259 |         hash = (hash + 1) % node_hash_size;
260 |     }
261 |     return -1;
262 | }
263 | 
264 | // Return position of a node in the node list of the specific type; if the node is not found, returns -1
265 | long long TypeSearchNode(char *node_str, long long type_Id)
266 | {
267 |     long long hash = GetNodeHash(node_str);
268 |     while (1)
269 |     {
270 |         if (type_node_hash[type_Id][hash] == -1) return -1;
271 |         if (!strcmp(node_str, type_node_list[type_Id][type_node_hash[type_Id][hash]].node_str)) return type_node_hash[type_Id][hash];
272 |         hash = (hash + 1) % node_hash_size;
273 |     }
274 |     return -1;
275 | }
276 | 
277 | // Return position of a node context pair in the node context list; if the node is not found, returns -1
278 | long long SearchNodeContextPair(long long node, long long context)
279 | {
280 |     long long hash = GetNodeContextHash(node_list[node].node_str, node_list[context].node_str);
281 |     long long hash_iter = 0;
282 |     long long cur_node, cur_context;
283 |     while(1)
284 |     {
285 |         if (node_context_hash[hash] == -1) return -1;
286 |         cur_node = node_context_list[node_context_hash[hash]].source;
287 |         cur_context = node_context_list[node_context_hash[hash]].target;
288 |         if (cur_node == node && cur_context == context) return node_context_hash[hash];
289 |         hash = (hash + 1) % node_context_hash_size;
290 |         hash_iter++;
291 |         if (hash_iter >= node_context_hash_size)
292 |         {
293 |             printf("The node context hash table is full!\n");
294 |             exit(1);
295 |         }
296 |     }
297 |     return -1;
298 | }
299 | 
300 | // Adds a node to the node list
301 | long long AddNodeToList(char *node_str)
302 | {
303 |     long long hash;
304 |     long long length = strlen(node_str) + 1;
305 |     if (length > MAX_STRING) length = MAX_STRING;
306 |     node_list[node_list_size].node_str = (char *)calloc(length, sizeof(char));
307 |     if (node_list[node_list_size].node_str == NULL)
308 |     {
309 |         printf("Memory allocation failed\n");
310 |         exit(1);
311 |     }
312 |     strcpy(node_list[node_list_size].node_str, node_str);
313 |     node_list[node_list_size].cn = 1;
314 |     node_list_size++;
315 |     // Reallocate memory if needed
316 |     if (node_list_size >= node_list_max_size)
317 |     {
318 |         node_list_max_size += 10000;
319 |         node_list = (struct Node *)realloc(node_list, node_list_max_size * sizeof(struct Node));
320 |         if (node_list == NULL)
321 |         {
322 |             printf("Memory allocation failed\n");
323 |             exit(1);
324 |         }
325 |     }
326 |     hash = GetNodeHash(node_str);
327 |     while (node_hash[hash] != -1) hash = (hash + 1) % node_hash_size;
328 |     node_hash[hash] = node_list_size - 1;
329 |     return node_list_size - 1;
330 | }
331 | 
332 | //Add node context pair to the node context list
333 | long long AddNodeContextToList(long long node, long long context)
334 | {
335 |     long long hash;
336 |     long long hash_iter = 0;
337 |     node_context_list[node_context_list_size].source = node;
338 |     node_context_list[node_context_list_size].target = context;
339 |     node_context_list[node_context_list_size].cn = 1;
340 |     node_context_list_size++;
341 |     if (node_context_list_size >= node_context_list_max_size)
342 |     {
343 |         node_context_list_max_size += 100 * node_list_size;
344 |         node_context_list = (struct Node_Context *)realloc(node_context_list, node_context_list_max_size * sizeof(struct Node_Context));
345 |         if (node_context_list == NULL)
346 |         {
347 |             printf("Memory allocation failed\n");
348 |             exit(1);
349 |         }
350 |     }
351 |     hash = GetNodeContextHash(node_list[node].node_str, node_list[context].node_str);
352 |     while (node_context_hash[hash] != -1)
353 |     {
354 |         hash = (hash + 1) % node_context_hash_size;
355 |         hash_iter++;
356 |         if (hash_iter >= node_context_hash_size)
357 |         {
358 |             printf("The node context hash table is full!\n");
359 |             exit(1);
360 |         }
361 |     }
362 |     node_context_hash[hash] = node_context_list_size - 1;
363 |     return node_context_list_size - 1;
364 | }
365 | 
366 | void LearnNodeListFromTrainFile()
367 | {
368 |     char node_str[MAX_STRING];
369 |     FILE *fin;
370 |     long long a, i;
371 |     for (a = 0; a < node_hash_size; a++) node_hash[a] = -1;
372 |     fin = fopen(train_file, "r");
373 |     if (fin == NULL)
374 |     {
375 |         printf("ERROR: training data file not found!\n");
376 |         exit(1);
377 |     }
378 |     node_list_size = 0;
379 |     printf("Reading nodes...\n");
380 |     while (1)
381 |     {
382 |         ReadNode(node_str, fin);
383 |         if (feof(fin)) break;
384 |         node_occur_num++;
385 |         if (node_occur_num % 100000 == 0)
386 |         {
387 |             printf("Read nodes: %lldK%c", node_occur_num / 1000, 13);
388 |             fflush(stdout);
389 |         }
390 |         i = SearchNode(node_str);
391 |         if (i == -1) AddNodeToList(node_str);
392 |         else node_list[i].cn++;
393 |     }
394 |     printf("Node list size: %lld \n", node_list_size);
395 |     printf("Node occurrence times in train file: %lld\n", node_occur_num);
396 |     fclose(fin);
397 | }
398 | 
399 | void GetNodeContextTable()
400 | {
401 |     char node_str[MAX_STRING];
402 |     long long a, b, c, node, context_node, walk_length = 0, walk_position = 0;
403 |     long long walk[MAX_WALK_LENGTH + 1];
404 |     long long line_num = 0;
405 |     long long end_flag;
406 |     FILE *fi = fopen(train_file, "r");
407 |     node_context_hash = (long long *)malloc(node_context_hash_size * sizeof(long long));
408 |     node_context_list_max_size = 1000 * node_list_size;
409 |     node_context_list = (struct Node_Context *)malloc(node_context_list_max_size * sizeof(struct Node_Context));
410 |     for (a = 0; a < node_context_hash_size; a++) node_context_hash[a] = -1;
411 |     node_context_list_size = 0;
412 |     printf("Collecting node context pair...\n");
413 |     start = clock();
414 |     while (1)
415 |     {
416 |         if (walk_length == 0)
417 |         {
418 |             while (1)
419 |             {
420 |                 end_flag = ReadNode(node_str, fi);
421 |                 if (feof(fi)) break;
422 |                 node = SearchNode(node_str);
423 |                 //if (node == -1) continue;
424 |                 walk[walk_length] = node;
425 |                 walk_length++;
426 |                 if (end_flag || walk_length >= MAX_WALK_LENGTH) break;
427 |             }
428 |             walk_position = 0;
429 |             line_num++;
430 |             if (line_num % 1000 == 0)
431 |             {
432 |                 printf("Processed lines: %lldK%c", line_num / 1000, 13);
433 |                 fflush(stdout);
434 |             }
435 |         }
436 |         if (feof(fi) && walk_length == 0) break;
437 |         //if (walk_length == 0) continue;
438 |         node = walk[walk_position]; //current node
439 |         for (a = 0; a < window * 2 + 1; a++)
440 |         {
441 |             if (a != window)
442 |             {
443 |                 c = walk_position - window + a;
444 |                 if (c < 0) continue;
445 |                 if (c >= walk_length) continue;
446 |                 context_node = walk[c];
447 |                 b = SearchNodeContextPair(node, context_node);
448 |                 if (b == -1) AddNodeContextToList(node, context_node);
449 |                 else node_context_list[b].cn++;
450 |             }
451 |         }
452 |         walk_position++;
453 |         if (walk_position >= walk_length) walk_length = 0;
454 |     }
455 |     finish = clock();
456 |     printf("Total time: %lf secs for collecting node context pairs\n", (double)(finish - start) / CLOCKS_PER_SEC);
457 |     printf("Node context pair number: %lld\n", node_context_list_size);
458 |     printf("----------------------------------------------------\n");
459 | }
460 | 
461 | void InitNet()
462 | {
463 |     long long a, b;
464 |     unsigned long long next_random = 1;
465 |     syn0 = (double *)malloc(node_list_size * layer1_size * sizeof(double));
466 |     if (syn0 == NULL)
467 |     {
468 |         printf("Memory allocation failed\n");
469 |         exit(1);
470 |     }
471 |     for (a = 0; a < node_list_size; a++)
472 |         for (b = 0; b < layer1_size; b++)
473 |         {
474 |             next_random = next_random * (unsigned long long)25214903917 + 11;
475 |             syn0[a * layer1_size + b] = (((next_random & 0xFFFF) / (double)65536) - 0.5) / layer1_size;
476 |         }
477 |     syn1neg = (double *)malloc(node_list_size * layer1_size * sizeof(double));
478 |     if (syn1neg == NULL)
479 |     {
480 |         printf("Memory allocation failed\n");
481 |         exit(1);
482 |     }
483 |     for (a = 0; a < node_list_size; a++)
484 |         for (b = 0; b < layer1_size; b++)
485 |             syn1neg[a * layer1_size + b] = 0;
486 |     InitUnigramTable();
487 | }
488 | 
489 | /* The alias sampling algorithm, which is used to sample an node context pair in O(1) time. */
490 | void InitAliasTable()
491 | {
492 |     long long k;
493 |     double sum = 0;
494 |     long long cur_small_block, cur_large_block;
495 |     long long num_small_block = 0, num_large_block = 0;
496 |     double *norm_prob;
497 |     long long *large_block;
498 |     long long *small_block;
499 |     alias = (long long *)malloc(node_context_list_size * sizeof(long long));
500 |     prob = (double *)malloc(node_context_list_size * sizeof(double));
501 |     if (alias == NULL || prob == NULL)
502 |     {
503 |         printf("Memory allocation failed\n");
504 |         exit(1);
505 |     }
506 |     norm_prob = (double*)malloc(node_context_list_size * sizeof(double));
507 |     large_block = (long long*)malloc(node_context_list_size * sizeof(long long));
508 |     small_block = (long long*)malloc(node_context_list_size * sizeof(long long));
509 |     if (norm_prob == NULL || large_block == NULL || small_block == NULL)
510 |     {
511 |         printf("Memory allocation failed\n");
512 |         exit(1);
513 |     }
514 |     for (k = 0; k != node_context_list_size; k++) sum += node_context_list[k].cn;
515 |     for (k = 0; k != node_context_list_size; k++) norm_prob[k] = (double)node_context_list[k].cn * node_context_list_size / sum;
516 |     for (k = node_context_list_size - 1; k >= 0; k--)
517 |     {
518 |         if (norm_prob[k]<1)
519 |             small_block[num_small_block++] = k;
520 |         else
521 |             large_block[num_large_block++] = k;
522 |     }
523 |     while (num_small_block && num_large_block)
524 |     {
525 |         cur_small_block = small_block[--num_small_block];
526 |         cur_large_block = large_block[--num_large_block];
527 |         prob[cur_small_block] = norm_prob[cur_small_block];
528 |         alias[cur_small_block] = cur_large_block;
529 |         norm_prob[cur_large_block] = norm_prob[cur_large_block] + norm_prob[cur_small_block] - 1;
530 |         if (norm_prob[cur_large_block] < 1)
531 |             small_block[num_small_block++] = cur_large_block;
532 |         else
533 |             large_block[num_large_block++] = cur_large_block;
534 |     }
535 |     while (num_large_block) prob[large_block[--num_large_block]] = 1;
536 |     while (num_small_block) prob[small_block[--num_small_block]] = 1;
537 |     free(norm_prob);
538 |     free(small_block);
539 |     free(large_block);
540 | }
541 | 
542 | long long SampleAPair(double rand_value1, double rand_value2)
543 | {
544 |     long long k = (long long)node_context_list_size * rand_value1;
545 |     return rand_value2 < prob[k] ? k : alias[k];
546 | }
547 | 
548 | void TrainModel()
549 | {
550 |     long a, b, c, d;
551 |     long long node, context_node, cur_pair;
552 |     long long count = 0, last_count = 0;
553 |     long long l1, l2, target, label;
554 |     long long type_Id;
555 |     unsigned long long next_random = 1;
556 |     double rand_num1, rand_num2;
557 |     double f, g;
558 |     double *neu1e = (double *)calloc(layer1_size, sizeof(double));
559 |     FILE *fp;
560 |     InitNet();
561 |     starting_alpha = alpha;
562 |     printf("Skip-Gram model with Negative Sampling:");
563 |     if (pp == 1) printf(" heterogeneous version\n");
564 |     else printf(" homogeneous version\n");
565 |     printf("Training file: %s\n", train_file);
566 |     printf("Samples: %lldM\n", total_samples / 1000000);
567 |     printf("Dimension: %lld\n", layer1_size);
568 |     printf("Initial Alpha: %f\n", alpha);
569 |     start = clock();
570 |     srand((unsigned)time(NULL));
571 |     while (1)
572 |     {
573 |         if (count >= total_samples) break;
574 |         if (count - last_count > 10000)
575 |         {
576 |             last_count = count;
577 |             printf("Alpha: %f Progress %.3lf%%%c", alpha, (double)count / (double)(total_samples + 1) * 100, 13);
578 |             fflush(stdout);
579 |             alpha = starting_alpha * (1 - (double)count / (double)(total_samples + 1));
580 |             if (alpha < starting_alpha * 0.0001) alpha = starting_alpha * 0.0001;
581 |         }
582 |         rand_num1 = rand() / (RAND_MAX * 1.0 + 1);
583 |         rand_num2 = rand() / (RAND_MAX * 1.0 + 1);
584 |         cur_pair = SampleAPair(rand_num1, rand_num2);
585 |         node = node_context_list[cur_pair].source;
586 |         context_node = node_context_list[cur_pair].target;
587 |         l1 = node * layer1_size;
588 |         for (c = 0; c < layer1_size; c++) neu1e[c] = 0;
589 | 
590 |         // NEGATIVE SAMPLING
591 |         for (d = 0; d < negative + 1; d++)
592 |         {
593 |             if (d == 0)
594 |             {
595 |                 target = context_node;
596 |                 label = 1;
597 |             }
598 |             else
599 |             {
600 |                 next_random = next_random * (unsigned long long)25214903917 + 11;
601 |                 if (pp == 1)
602 |                 {
603 |                     type_Id = GetTypeId(node_list[context_node].node_str[0]);
604 |                     target = type_tables[type_Id][(next_random >> 16) % type_counts[type_Id]];
605 |                 }
606 |                 else
607 |                     target = table[(next_random >> 16) % table_size];
608 |                 if (target == context_node) continue;
609 |                 label = 0;
610 |             }
611 |             l2 = target * layer1_size;
612 |             f = 0;
613 |             for (c = 0; c < layer1_size; c++) f += syn0[c + l1] * syn1neg[c + l2];
614 |             if (f > MAX_EXP) f = 1;
615 |             else if (f < -MAX_EXP) f = 0;
616 |             else f = expTable[(int)((f + MAX_EXP) * (EXP_TABLE_SIZE / MAX_EXP / 2))];
617 |             g = (label - f) * alpha;
618 |             // Propagate errors output -> hidden
619 |             for (c = 0; c < layer1_size; c++) neu1e[c] += g * syn1neg[c + l2];
620 |             // Learn weights hidden -> output
621 |             for (c = 0; c < layer1_size; c++) syn1neg[c + l2] += g * syn0[c + l1];
622 |         }
623 |         // Learn weights input -> hidden
624 |         for (c = 0; c < layer1_size; c++) syn0[c + l1] += neu1e[c];
625 |         count++;
626 |     }
627 |     finish = clock();
628 |     printf("Total time: %lf secs for learning node embeddings\n", (double)(finish - start) / CLOCKS_PER_SEC);
629 |     printf("----------------------------------------------------\n");
630 |     fp = fopen(output_file, "w");
631 |     // Save the learned node representations
632 |     for (a = 0; a < node_list_size; a++)
633 |     {
634 |         if (node_list[a].node_str[0] != prefixes[obj_type]) continue;
635 |         fprintf(fp, "%s", node_list[a].node_str);
636 |         for (b = 0; b < layer1_size; b++) fprintf(fp, " %lf", syn0[a * layer1_size + b]);
637 |         fprintf(fp, "\n");
638 |     }
639 |     free(neu1e);
640 |     fclose(fp);
641 | }
642 | 
643 | int ArgPos(char *str, int argc, char **argv)
644 | {
645 |     int a;
646 |     for (a = 1; a < argc; a++) if (!strcmp(str, argv[a]))
647 |         {
648 |             if (a == argc - 1)
649 |             {
650 |                 printf("Argument missing for %s\n", str);
651 |                 exit(1);
652 |             }
653 |             return a;
654 |         }
655 |     return -1;
656 | }
657 | 
658 | int main(int argc, char **argv)
659 | {
660 |     int i;
661 |     if (argc == 1)
662 |     {
663 |         printf("---randomWalk2vec---\n");
664 |         printf("---The code and following instructions are built upon word2vec.c by Mikolov et al.---\n\n");
665 |         printf("Options:\n");
666 |         printf("Parameters for training:\n");
667 |         printf("\t-train <file>\n");
668 |         printf("\t\tUse random walk sequences from <file> to train the model\n");
669 |         printf("\t-output <file>\n");
670 |         printf("\t\tUse <file> to save the learned node vector-format representations\n");
671 |         printf("\t-size <int>\n");
672 |         printf("\t\tSet the dimension of learned node representations; default is 128\n");
673 |         printf("\t-window <int>\n");
674 |         printf("\t\tSet the window size for collecting node context pairs; default is 5\n");
675 |         printf("\t-pp <int>\n");
676 |         printf("\t\tUse Heterogeneous Skip-Gram model (the ++ version) or not; default is 1 (++ version); otherwise, use 0 (Homogeneous Skip-Gram)\n");
677 |         printf("\t-prefixes <string>\n");
678 |         printf("\t\tPrefixes of node Ids for specifying node types, e.g., ap with a for author and p for paper\n");
679 |         printf("\t-objtype\n");
680 |         printf("\t\tThe index of the objective node type in the prefixes list, for which representations to be learned\n");
681 |         printf("\t-alpha <float>\n");
682 |         printf("\t\tSet the starting learning rate; default is 0.025\n");
683 |         printf("\t-negative <int>\n");
684 |         printf("\t\tNumber of negative examples; default is 5, common values are 3 - 10\n");
685 |         printf("\t-samples <int>\n");
686 |         printf("\t\tSet the number of iterations for stochastic gradient descent as <int> Million; default is 100\n");
687 |         return 0;
688 |     }
689 |     if ((i = ArgPos((char *)"-size", argc, argv)) > 0) layer1_size = atoi(argv[i + 1]);
690 |     if ((i = ArgPos((char *)"-train", argc, argv)) > 0) strcpy(train_file, argv[i + 1]);
691 |     if ((i = ArgPos((char *)"-alpha", argc, argv)) > 0) alpha = atof(argv[i + 1]);
692 |     if ((i = ArgPos((char *)"-output", argc, argv)) > 0) strcpy(output_file, argv[i + 1]);
693 |     if ((i = ArgPos((char *)"-window", argc, argv)) > 0) window = atoi(argv[i + 1]);
694 |     if ((i = ArgPos((char *)"-pp", argc, argv)) > 0) pp = atoi(argv[i + 1]);
695 |     if ((i = ArgPos((char *)"-prefixes", argc, argv)) > 0) strcpy(prefixes, argv[i + 1]);
696 |     if ((i = ArgPos((char *)"-objtype", argc, argv)) > 0) obj_type = atoi(argv[i + 1]);
697 |     if ((i = ArgPos((char *)"-negative", argc, argv)) > 0) negative = atoi(argv[i + 1]);
698 |     if ((i = ArgPos((char *)"-samples", argc, argv)) >0) total_samples = atoi(argv[i + 1]);
699 |     total_samples = total_samples * 1000000;
700 |     type_num = strlen(prefixes);
701 |     printf("Number of node types: %lld\n", type_num);
702 |     node_list = (struct Node *)calloc(node_list_max_size, sizeof(struct Node));
703 |     node_hash = (long long *)calloc(node_hash_size, sizeof(long long));
704 |     expTable = (double *)malloc((EXP_TABLE_SIZE + 1) * sizeof(double));
705 |     if (node_list == NULL || node_hash == NULL || expTable == NULL)
706 |     {
707 |         printf("Memory allocation failed\n");
708 |         exit(1);
709 |     }
710 |     for (i = 0; i < EXP_TABLE_SIZE; i++)
711 |     {
712 |         expTable[i] = exp((i / (double)EXP_TABLE_SIZE * 2 - 1) * MAX_EXP); // Precompute the exp() table
713 |         expTable[i] = expTable[i] / (expTable[i] + 1);                   // Precompute f(x) = x / (x + 1)
714 |     }
715 |     LearnNodeListFromTrainFile();
716 |     GetNodeContextTable();
717 |     InitAliasTable();
718 |     TrainModel();
719 |     return 0;
720 | }
721 | 


--------------------------------------------------------------------------------