edu.umass.cs.mallet.projects.seg_plus_coref.condclust.cluster

Class ConditionalClusterer

java.lang.Object

edu.umass.cs.mallet.projects.seg_plus_coref.condclust.cluster.ConditionalClusterer

public class ConditionalClusterer
extends Object

Trains the conditional clusterer to predict "yes" or "no" for a NodeClusterPair; i.e. does this nodes belong in this cluster?

Constructor Summary

Constructors

Constructor and Description
ConditionalClusterer(Pipe _pipe, Classifier _classifier)
ConditionalClusterer(Pipe _pipe, Classifier _classifier, double _threshold)

Method Summary

Modifier and Type	Method and Description
Collection	cluster(ArrayList _nodes, Collection trueClustering) Greedily cluster by adding node to Clustering that is (a) closest to an existing cluster, or (b) farthest from all existing clusters (i.e.
Collection	clusterPapersAndVenues(ArrayList _papers, ArrayList _venues, Collection paperTrueClustering, Collection venueTrueClustering, Classifier paperClusterClassifier, Classifier venueClusterClassifier, Random r) Cluster papers and venues jointly.
Collection	clusterRandom(ArrayList _nodes, Collection trueClustering, Random r) Pick a random ordering to cluster nodes, using the learned classifier to make yes/no decisions.
Classifier	getClassifier()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ConditionalClusterer

public ConditionalClusterer(Pipe _pipe,
                            Classifier _classifier,
                            double _threshold)

ConditionalClusterer

public ConditionalClusterer(Pipe _pipe,
                            Classifier _classifier)

Method Detail

getClassifier

public Classifier getClassifier()

clusterPapersAndVenues

public Collection clusterPapersAndVenues(ArrayList _papers,
                                         ArrayList _venues,
                                         Collection paperTrueClustering,
                                         Collection venueTrueClustering,
                                         Classifier paperClusterClassifier,
                                         Classifier venueClusterClassifier,
                                         Random r)

Cluster papers and venues jointly.

cluster

public Collection cluster(ArrayList _nodes,
                          Collection trueClustering)

Greedily cluster by adding node to Clustering that is (a) closest to an existing cluster, or (b) farthest from all existing clusters (i.e. closest to being a new cluster)

clusterRandom

public Collection clusterRandom(ArrayList _nodes,
                                Collection trueClustering,
                                Random r)

Pick a random ordering to cluster nodes, using the learned classifier to make yes/no decisions.