Added configuratino options for clustering coefficient

Author: ArnauPrat

src/main/java/ldbc/snb/datagen/generator/ClusteringKnowsGenerator.java

@@ -7,6 +7,7 @@
 import ldbc.snb.datagen.objects.Knows;
 import ldbc.snb.datagen.objects.Person;
 import ldbc.snb.datagen.util.RandomGeneratorFarm;
+import org.apache.hadoop.conf.Configuration;
 
 import java.util.*;
 
@@ -18,6 +19,7 @@ public class ClusteringKnowsGenerator implements KnowsGenerator {
     Random rand;
     private ArrayList<Float> percentages = null;
     private int stepIndex = 0;
+    private float targetCC = 0.0f;
 
     private class PersonInfo {
         public int index_;
@@ -333,7 +335,7 @@ boolean improveCC(ClusteringInfo cInfo, ArrayList<Community> communities) {
         Community c = filtered.get(index);
         c.p_ = Math.max(c.p_ + 0.01f, 1.0f);
         cInfo.sumProbs+=0.01;
-        estimateCCCommunity(cInfo, c, c.p_ );
+        estimateCCCommunity(cInfo, c, c.p_);
         return true;
     }
 
@@ -421,7 +423,6 @@ public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float>
         rand.setSeed(seed);
         this.percentages = percentages;
         this.stepIndex = step_index;
-        float targetCC = 0.15f;
 
 
         ArrayList<Community>  communities = generateCommunities(persons);
@@ -457,4 +458,10 @@ public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float>
         }
         System.out.println("Number of persons with less degree than expected: "+count);
     }
+
+    public void initialize( Configuration conf ) {
+        targetCC = conf.getFloat("ldbc.snb.datagen.generator.ClusteringKnowsGenerator.clusteringCoefficient", 0.1f);
+        System.out.println("Initialized clustering coefficient to "+targetCC);
+        targetCC /= 2.0f;
+    }
 }

src/main/java/ldbc/snb/datagen/generator/DistanceKnowsGenerator.java

@@ -4,6 +4,7 @@
 import ldbc.snb.datagen.objects.Knows;
 import ldbc.snb.datagen.objects.Person;
 import ldbc.snb.datagen.util.RandomGeneratorFarm;
+import org.apache.hadoop.conf.Configuration;
 
 import java.util.ArrayList;
 
@@ -30,6 +31,10 @@ public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float>
         }
     }
 
+    public void initialize( Configuration conf ) {
+
+    }
+
     boolean know( Person personA, Person personB, int dist, ArrayList<Float> percentages, int step_index ) {
         if( personA.knows().size() >= Knows.target_edges( personA, percentages, step_index) ||
             personB.knows().size() >= Knows.target_edges( personB, percentages, step_index) ) return false;
@@ -42,4 +47,5 @@ boolean know( Person personA, Person personB, int dist, ArrayList<Float> percent
     }
 
 
+
 }

src/main/java/ldbc/snb/datagen/generator/KnowsGenerator.java

@@ -1,6 +1,7 @@
 package ldbc.snb.datagen.generator;
 
 import ldbc.snb.datagen.objects.Person;
+import org.apache.hadoop.conf.Configuration;
 
 import java.util.ArrayList;
 
@@ -9,4 +10,6 @@
  */
 public interface KnowsGenerator {
     public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float> percentages, int step_index );
+
+    public void initialize( Configuration  conf );
 }

src/main/java/ldbc/snb/datagen/generator/RandomKnowsGenerator.java

@@ -2,6 +2,7 @@
 
 import ldbc.snb.datagen.objects.Knows;
 import ldbc.snb.datagen.objects.Person;
+import org.apache.hadoop.conf.Configuration;
 
 import java.util.ArrayList;
 import java.util.Collections;
@@ -51,4 +52,8 @@ public void generateKnows( ArrayList<Person> persons, int seed, ArrayList<Float>
             }
         }
     }
+
+    public void initialize( Configuration conf ) {
+
+    }
 }

src/main/java/ldbc/snb/datagen/hadoop/HadoopKnowsGenerator.java

@@ -37,6 +37,7 @@ protected void setup(Context context) {
             this.conf = context.getConfiguration();
             try {
                 this.knowsGenerator = (KnowsGenerator) Class.forName(conf.get("knowsGeneratorName")).newInstance();
+                this.knowsGenerator.initialize(conf);
             }catch(Exception e) {
                 System.out.println(e.getMessage());
             }
@@ -80,7 +81,7 @@ public void reduce(BlockKey key, Iterable<Person> valueSet,Context context)
 
 
     public HadoopKnowsGenerator( Configuration conf, String preKeySetterName, String postKeySetterName, ArrayList<Float> percentages, int step_index, String knowsGeneratorName  ) {
-        this.conf = conf;
+        this.conf = new Configuration(conf);
         this.preKeySetterName = preKeySetterName;
         this.postKeySetterName = postKeySetterName;
         this.percentages = percentages;