export graphs and vertices as csv #1

commands.sh

@@ -1,41 +1,42 @@
 #!/bin/sh
 
-java -Xmx80g -jar graphlod-0.1.jar --excludedNamespaces \
+java -Xmx80g -jar graphlod-0.1.jar --name drugbank --excludedNamespaces \
     "http://www4.wiwiss.fu-berlin.de/drugbank/resource/drugtype/" \
     "http://www4.wiwiss.fu-berlin.de/drugbank/resource/references/" \
     --skipChromatic \
     --namespace "http://www4.wiwiss.fu-berlin.de/drugbank/" \
     /data/graphlod/drugbank/drugbank.nt | tee drugbank.txt
 
-java -Xmx80g -jar graphlod-0.1.jar --skipChromatic \
+java -Xmx80g -jar graphlod-0.1.jar  --name dailymed --skipChromatic \
     --namespace "http://www4.wiwiss.fu-berlin.de/dailymed/" \
     /data/graphlod/dailymed/dailymed_dump.nt | tee dailymed_dump.txt
 
-java -Xmx80g -jar graphlod-0.1.jar  --skipChromatic --excludedNamespaces \
+java -Xmx80g -jar graphlod-0.1.jar --name diseasome --skipChromatic --excludedNamespaces \
     "http://www4.wiwiss.fu-berlin.de/diseasome/resource/diseaseClass/" \
     --namespace "http://www4.wiwiss.fu-berlin.de/diseasome/" \
     /data/graphlod/diseasome/diseasome.nt | tee diseasome.txt
 
-java -Xmx80g -jar graphlod-0.1.jar --skipChromatic \
+java -Xmx80g -jar graphlod-0.1.jar --name dbpedia_person --skipChromatic \
     --namespace "http://dbpedia.org/resource" \
     /data/graphlod/dbpedia/persondata_en.nt | tee dbpedia_persondata.txt
 
-java -Xmx80g -jar graphlod-0.1.jar --skipChromatic  \
+java -Xmx80g -jar graphlod-0.1.jar --name dbpedia_geo_coordinate --skipChromatic  \
     --namespace "http://dbpedia.org/resource" \
     /data/graphlod/dbpedia/geo_coordinates_en.nt | tee geo_coordinate.txt
 
-java -Xmx80g -jar graphlod-0.1.jar --skipChromatic  \
+java -Xmx80g -jar graphlod-0.1.jar --name dbpedia_homepages --skipChromatic  \
     --namespace "http://dbpedia.org/resource" \
     /data/graphlod/dbpedia/homepages_en.nt | tee dbpedia_homepages.txt
 
 # fix mapping: sed 's/"\.$/" \./' mappingbased_properties_en.nt > mappingbased_properties_en_fixed.nt
 
-java -Xmx100g -jar graphlod-0.1.jar --skipChromatic \
+java -Xmx100g -jar graphlod-0.1.jar --name dbpedia_mapping --skipChromatic \
     --namespace "http://dbpedia.org/resource" \
     mappingbased_properties_en_fixed.nt | tee dbpedia_mapping.txt
 
-java -Xmx80g -jar graphlod-0.1.jar --skipChromatic \
+java -Xmx80g -jar graphlod-0.1.jar --name linkedgeodata --skipChromatic \
     /data/graphlod/linkedgeodata/2013-04-29-{Ae*,C*,E*,Mili*,H*,P*,S*,T*} \
     --namespace "http://linkedgeodata.org/" \
     | tee linkedgeodata.txt
 
+zip result.txt *.txt *.csv

pom.xml

@@ -55,6 +55,11 @@
             <artifactId>commons-lang3</artifactId>
             <version>3.3.2</version>
         </dependency>
+        <dependency>
+            <groupId>org.apache.commons</groupId>
+            <artifactId>commons-csv</artifactId>
+            <version>1.0</version>
+        </dependency>
         <dependency>
             <groupId>com.google.guava</groupId>
             <artifactId>guava</artifactId>

src/graphlod/CollectionUtils.java

@@ -31,7 +31,7 @@ public static <T extends Comparable<T>> T min(Collection<T> collection) {
         }
         T result = collection.iterator().next();
         for (T element : collection) {
-            result = result.compareTo(element) > 0 ? result : element;
+            result = result.compareTo(element) < 0 ? result : element;
         }
         return result;
     }
@@ -42,7 +42,7 @@ public static <T extends Comparable<T>> T max(Collection<T> collection) {
         }
         T result = collection.iterator().next();
         for (T element : collection) {
-            result = result.compareTo(element) < 0 ? result : element;
+            result = result.compareTo(element) > 0 ? result : element;
         }
         return result;
     }

src/graphlod/Dataset.java

@@ -72,7 +72,7 @@ private void readTriples(NxParser nxp) {
             }
 
             if (subjectUri.equals(objectUri)) {
-                continue; // TODO: why that?
+                continue;
             }
 
             if (propertyUri.equals("http://www.w3.org/1999/02/22-rdf-syntax-ns#type")) {

src/graphlod/GraphCsvOutput.java

@@ -0,0 +1,53 @@
+package graphlod;
+
+
+import java.io.IOException;
+import java.io.Writer;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.List;
+import java.util.Set;
+
+import org.apache.commons.csv.CSVFormat;
+import org.apache.commons.csv.CSVPrinter;
+
+import com.google.common.base.Charsets;
+import com.google.common.base.Verify;
+
+public class GraphCsvOutput {
+
+	private final CSVPrinter writer;
+	private final int maxSizeForDiameter;
+
+	public GraphCsvOutput(String name, int maxSizeForDiameter) {
+		this.maxSizeForDiameter = maxSizeForDiameter;
+		Writer out;
+		try {
+			Path path = Paths.get(name + "_graphs.csv");
+			out = Files.newBufferedWriter(path, Charsets.UTF_8);
+			writer = CSVFormat.DEFAULT.withHeader("graph", "vertices", "edges", "diameter", "avgindegree", "maxindegree", "avgoutdegree", "maxoutdegree").print(out);
+		} catch (IOException e) {
+			throw new RuntimeException(e);
+		}
+	}
+
+	public void writeGraph(GraphFeatures graph) {
+		double diameter = graph.getVertexCount() < maxSizeForDiameter ? graph.getDiameter() : -1;
+		try {
+			writer.printRecord(graph.getId(), graph.getVertexCount(), graph.getEdgeCount(), diameter,
+					CollectionUtils.average(graph.getIndegrees()), CollectionUtils.max(graph.getIndegrees()),
+					CollectionUtils.average(graph.getOutdegrees()), CollectionUtils.max(graph.getOutdegrees()));
+		} catch (IOException e) {
+			e.printStackTrace();
+		}
+	}
+	public void close() {
+		try {
+			writer.close();
+		} catch (IOException e) {
+			throw new RuntimeException(e);
+		}
+	}
+
+}

src/graphlod/GraphFeatures.java

@@ -30,8 +30,10 @@ public class GraphFeatures {
 	private Set<String> vertices;
 	private final Set<DefaultEdge> edges;
 	private AsUndirectedGraph<String, DefaultEdge> undirectedG;
+	private String id;
 
-	public GraphFeatures(DirectedGraph<String, DefaultEdge> graph) {
+	public GraphFeatures(String id, DirectedGraph<String, DefaultEdge> graph) {
+		this.id = id;
 		this.graph = graph;
 		this.vertices = this.graph.vertexSet();
 		this.edges = this.graph.edgeSet();
@@ -79,7 +81,7 @@ public List<GraphFeatures> getConnectedSubGraphFeatures(float minSize) {
 			return Collections.emptyList();
 		}
 		List<GraphFeatures> connectedSubgraphFeatures = new ArrayList<>();
-
+		int i = 0;
 		for (Set<String> set : sets) {
 			if (set.size() < minSize) {
 				continue;
@@ -94,7 +96,8 @@ public List<GraphFeatures> getConnectedSubGraphFeatures(float minSize) {
 					subgraph.addEdge(vertex, (String) edge.getTarget(), edge);
 				}
 			}
-			connectedSubgraphFeatures.add(new GraphFeatures(subgraph));
+			connectedSubgraphFeatures.add(new GraphFeatures("subgraph" + i, subgraph));
+			i++;
 		}
 		Collections.sort(connectedSubgraphFeatures, new Comparator<GraphFeatures>() {
 			@Override
@@ -132,6 +135,13 @@ public List<Integer> getIndegrees() {
 		return this.indegrees;
 	}
 
+	public List<Degree> getIndegrees2() {
+		if(this.indegrees2 == null) {
+			getIndegrees();
+		}
+		return this.indegrees2;
+	}
+
 	public List<Integer> getOutdegrees() {
 		if (this.outdegrees == null) {
 			this.outdegrees = new ArrayList<>();
@@ -145,6 +155,14 @@ public List<Integer> getOutdegrees() {
 		return this.outdegrees;
 	}
 
+
+	public List<Degree> getOutdegrees2() {
+		if(this.outdegrees2 == null) {
+			getOutdegrees();
+		}
+		return this.outdegrees2;
+	}
+
 	public ArrayList<Integer> getEdgeCounts() {
 		ArrayList<Integer> edgeCounts = new ArrayList<>();
 		for (String vertex : this.vertices) {
@@ -165,6 +183,14 @@ public int getChromaticNumber() {
 		return ChromaticNumber.findGreedyChromaticNumber(this.undirectedG);
 	}
 
+	public Set<String> getVertices() {
+		return vertices;
+	}
+
+	public String getId() {
+		return id;
+	}
+
 	static class Degree implements Comparable<Degree> {
 		public String vertex;
 		public int degree;

src/graphlod/GraphLOD.java

@@ -28,11 +28,17 @@ public class GraphLOD {
 	private static final Logger logger = Logger.getLogger(GraphLOD.class);
 	public static final int MAX_SIZE_FOR_DIAMETER = 500;
 
-	public GraphLOD(Collection<String> datasetFiles, boolean skipChromaticNumber, String namespace, Collection<String> excludedNamespaces, float minImportantSubgraphSize, int importantDegreeCount) {
+	public GraphCsvOutput graphCsvOutput;
+	public VertexCsvOutput vertexCsvOutput;
+
+	public GraphLOD(String name, Collection<String> datasetFiles, boolean skipChromaticNumber, String namespace, Collection<String> excludedNamespaces, float minImportantSubgraphSize, int importantDegreeCount) {
+		graphCsvOutput = new GraphCsvOutput(name, MAX_SIZE_FOR_DIAMETER);
+		vertexCsvOutput = new VertexCsvOutput(name);
+
 		Stopwatch sw = Stopwatch.createStarted();
 		Dataset dataset = Dataset.fromFiles(datasetFiles, namespace, excludedNamespaces);
 
-		GraphFeatures graphFeatures = new GraphFeatures(dataset.getGraph());
+		GraphFeatures graphFeatures = new GraphFeatures("main_graph", dataset.getGraph());
 
 		System.out.println("Loading the dataset took " + sw + " to execute.");
 
@@ -85,6 +91,7 @@ public GraphLOD(Collection<String> datasetFiles, boolean skipChromaticNumber, St
 				System.out.printf("Subgraph: %s vertices\n", subGraph.getVertexCount());
 				analyzeConnectedGraph(subGraph, importantDegreeCount);
 			}
+
 			System.out.println("Analysing the subgraphs took " + sw + " to execute.");
 		}
 
@@ -109,6 +116,8 @@ public GraphLOD(Collection<String> datasetFiles, boolean skipChromaticNumber, St
 			System.out.println("Chromatic Number: " + cN);
 			System.out.println("Getting the Chromatic Number took " + sw + " to execute.");
 		}
+		graphCsvOutput.close();
+		vertexCsvOutput.close();
 	}
 
 	private void analyzeConnectedGraph(GraphFeatures graph, int importantDegreeCount) {
@@ -118,6 +127,8 @@ private void analyzeConnectedGraph(GraphFeatures graph, int importantDegreeCount
 		} else {
 			System.out.println("\tGraph too big to show diameter");
 		}
+		graphCsvOutput.writeGraph(graph);
+		vertexCsvOutput.writeGraph(graph);
 
 		System.out.println("\thighest indegrees:");
 		System.out.println("\t\t" + StringUtils.join(graph.maxInDegrees(importantDegreeCount), "\n\t\t"));
@@ -133,6 +144,7 @@ public static void main(final String[] args) {
 		ArgumentParser parser = ArgumentParsers.newArgumentParser("GraphLOD")
 				.defaultHelp(true).description("calculates graph features.");
 		parser.addArgument("dataset").nargs("+").setDefault(Arrays.asList(DEFAULT_DATASET_LOCATION));
+		parser.addArgument("--name").type(String.class).setDefault("");
 		parser.addArgument("--namespace").type(String.class).setDefault("");
 		parser.addArgument("--excludedNamespaces").nargs("*").setDefault(Collections.emptyList());
 		parser.addArgument("--skipChromatic").action(Arguments.storeTrue());
@@ -145,14 +157,20 @@ public static void main(final String[] args) {
 			parser.handleError(e);
 			System.exit(1);
 		}
+
 		List<String> dataset = result.getList("dataset");
+		String name = result.getString("name");
+		if(name.isEmpty()) {
+			name = dataset.get(0);
+		}
 		String namespace = result.getString("namespace");
 		List<String> excludedNamespaces = result.getList("excludedNamespaces");
 		boolean skipChromatic = result.getBoolean("skipChromatic");
 		int minImportantSubgraphSize = result.getInt("minImportantSubgraphSize");
 		int importantDegreeCount = result.getInt("importantDegreeCount");
 
 		System.out.println("reading: " + dataset);
+		System.out.println("name: " + name);
 		System.out.println("namespace: " + namespace);
 		System.out.println("skip chromatic: " + skipChromatic);
 		System.out.println("excluded namespaces: " + excludedNamespaces);
@@ -162,7 +180,7 @@ public static void main(final String[] args) {
 
 		Locale.setDefault(Locale.US);
 
-		new GraphLOD(dataset, skipChromatic, namespace, excludedNamespaces, minImportantSubgraphSize, importantDegreeCount);
+		new GraphLOD(name, dataset, skipChromatic, namespace, excludedNamespaces, minImportantSubgraphSize, importantDegreeCount);
 	}
 
 }

src/graphlod/VertexCsvOutput.java

@@ -0,0 +1,55 @@
+package graphlod;
+
+
+import java.io.IOException;
+import java.io.Writer;
+import java.nio.file.Files;
+import java.nio.file.Path;
+import java.nio.file.Paths;
+import java.util.List;
+
+import org.apache.commons.csv.CSVFormat;
+import org.apache.commons.csv.CSVPrinter;
+
+import com.google.common.base.Charsets;
+import com.google.common.base.Verify;
+
+public class VertexCsvOutput {
+
+	CSVPrinter writer;
+
+	public VertexCsvOutput(String name) {
+		Writer out;
+		try {
+			Path path = Paths.get(name + "_vertices.csv");
+			out = Files.newBufferedWriter(path, Charsets.UTF_8);
+			writer = CSVFormat.DEFAULT.withHeader("graph", "vertex", "indegree", "outdegree").print(out);
+		} catch (IOException e) {
+			throw new RuntimeException(e);
+		}
+	}
+
+	public void writeGraph(GraphFeatures graph) {
+		List<GraphFeatures.Degree> inDegrees = graph.getIndegrees2();
+		List<GraphFeatures.Degree> outDegrees = graph.getIndegrees2();
+
+		for (int i = 0; i < inDegrees.size(); i++) {
+			GraphFeatures.Degree in = inDegrees.get(i);
+			GraphFeatures.Degree out = outDegrees.get(i);
+			Verify.verify(in.vertex.equals(out.vertex));
+			try {
+				writer.printRecord(graph.getId(), in.vertex, in.degree, out.degree);
+			} catch (IOException e) {
+				throw new RuntimeException(e);
+			}
+		}
+	}
+	public void close() {
+		try {
+			writer.close();
+		} catch (IOException e) {
+			throw new RuntimeException(e);
+		}
+	}
+
+}

test/graphlod/CollectionUtilsTest.java

@@ -1,11 +1,13 @@
 package graphlod;
 
+import java.util.Arrays;
 import java.util.Collection;
 
 import org.junit.Test;
 
 import com.google.common.collect.Lists;
 
+import static org.hamcrest.Matchers.is;
 import static org.junit.Assert.*;
 import static org.hamcrest.Matchers.contains;
 import static org.hamcrest.Matchers.containsInAnyOrder;
@@ -16,4 +18,17 @@ public class CollectionUtilsTest {
 	public void testMaxValues() throws Exception {
 		assertThat(CollectionUtils.maxValues(Lists.newArrayList(5,2,9,2), 2), contains(5,9));
 	}
+
+	@Test
+	public void testMax() {
+		assertThat(CollectionUtils.max(Arrays.asList(1, 5, 3)), is(5));
+	}
+	@Test
+	public void testMin() {
+		assertThat(CollectionUtils.min(Arrays.asList(1, 5, 3)), is(1));
+	}
+	@Test
+	public void testAvg() {
+		assertThat(CollectionUtils.average(Arrays.asList(1, 5, 3)), is(3.0));
+	}
 }