org.carrot2.output.metrics
Class ContaminationMetric

java.lang.Object
  org.carrot2.output.metrics.ContaminationMetric

All Implemented Interfaces:

IClusteringMetric

public class ContaminationMetric
extends Object

Computes cluster contamination. If a cluster groups documents found in the same Document.PARTITIONS, its contamination is 0. If a cluster groups an equally distributed mix of all partitions, its contamination is 1.0. For a full definition, please see section 4.4.1 of this work.

Contamination is calculated for top-level clusters only, taking into account documents from the cluster and all subclusters. Finally, contamination will be calculated only if all input documents have non-blank Document.PARTITIONSs.

Field Summary

List<Cluster>	clusters
static String	CONTAMINATION Key for the contamination value of a cluster.
List<Document>	documents
boolean	enabled Calculate contamination metric.
String	partitionIdFieldName Partition id field name.
double	weightedAverageContamination Average contamination of the whole cluster set, weighted by the size of cluster.

Constructor Summary

ContaminationMetric()

Method Summary

void	calculate() Triggers calculation of the metric.
boolean	isEnabled() Return true if this metric should be calculated.

Methods inherited from class java.lang.Object

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

Field Detail

CONTAMINATION

public static final String CONTAMINATION

Key for the contamination value of a cluster.

See Also:

Constant Field Values

weightedAverageContamination

public double weightedAverageContamination

Average contamination of the whole cluster set, weighted by the size of cluster.

enabled

public boolean enabled

Calculate contamination metric.

documents

public List<Document> documents

clusters

public List<Cluster> clusters

partitionIdFieldName

public String partitionIdFieldName

Partition id field name.

Constructor Detail

ContaminationMetric

public ContaminationMetric()

Method Detail

calculate

public void calculate()

Description copied from interface: IClusteringMetric

Triggers calculation of the metric. All Processing Input attributes will have been bound before a call to this method.

isEnabled

public boolean isEnabled()

Description copied from interface: IClusteringMetric

Return true if this metric should be calculated.