netkit.classifiers.relational
Class ClassDistribRelNeighbor

java.lang.Object
  netkit.classifiers.ClassifierImp
      netkit.classifiers.relational.NetworkClassifierImp
          netkit.classifiers.relational.ClassDistribRelNeighbor

All Implemented Interfaces:

Classifier, NetworkClassifier, Configurable

public final class ClassDistribRelNeighbor
extends NetworkClassifierImp

The Class Distributional Relational Neighbor (ClassDistributRelNeighbor) classifier works by creating a 'prototypical' class vector for each class of node and then estimating a label for a new node by calculating how near that new node is to each of these 'class reference vectors'.

Properties:

• distance: the distance function to use. default=cosine. See distance.properties
• useintrinsic: whether to use intrinsic attributes. default=false
• aggregation: what attributes to aggregate on. default=classOnly
• aggregators: what aggregators to use. default=ratio

Author:

Sofus A. Macskassy (sofmac@gmail.com)

Nested Class Summary

Nested classes/interfaces inherited from class netkit.classifiers.relational.NetworkClassifierImp

NetworkClassifierImp.Aggregation

Field Summary

Fields inherited from class netkit.classifiers.relational.NetworkClassifierImp

aggFactory, aggregation, aggregators, aggTypes, dynamicAggregators, prior

Fields inherited from class netkit.classifiers.ClassifierImp

attribute, classPrior, clsIdx, graph, keyIndex, logger, nodeType, right, tmpVector, useIntrinsic, vectorClsIdx

Constructor Summary

ClassDistribRelNeighbor()

Method Summary

void	configure(Configuration config) Configure this classifier object.
boolean	doEstimate(Node node, double[] result) Estimate how near this node's neighborhood is to each of the class vectors using a user-specified distance function (cosine by default) and normalize to produce a pseudo distribution.
Configuration	getDefaultConfiguration() Get the detault configuration of using a cosine distance function, and aggregating only on the class attribute using the ratio aggregator.
java.lang.String	getDescription()
java.lang.String	getName()
java.lang.String	getShortName()
void	induceModel(Graph graph, DataSplit split) Induce the cdRN model by finding the 'prototypical' neighborhood for each class of nodes.
java.lang.String	toString()

Methods inherited from class netkit.classifiers.relational.NetworkClassifierImp

classify, estimate, estimate, estimate, estimate, generateAggregators, getAttributeNames, includeClassAttribute, initializeRun, makeVector

Methods inherited from class netkit.classifiers.ClassifierImp

addListener, classify, classify, clearListeners, estimate, estimate, getLogger, getNofifyListeners, notifyListeners, notifyListeners, removeListener, reset, setNofityListeners

Methods inherited from class java.lang.Object

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

Methods inherited from interface netkit.classifiers.Classifier

addListener, classify, classify, clearListeners, estimate, estimate, getLogger, getNofifyListeners, notifyListeners, notifyListeners, removeListener, reset, setNofityListeners

Constructor Detail

ClassDistribRelNeighbor

public ClassDistribRelNeighbor()

Method Detail

getDefaultConfiguration

public Configuration getDefaultConfiguration()

Get the detault configuration of using a cosine distance function, and aggregating only on the class attribute using the ratio aggregator. This is in addition to any defaults set by the superclass.

Specified by:

getDefaultConfiguration in interface Configurable

Overrides:

getDefaultConfiguration in class NetworkClassifierImp

Returns:

a Configuration object

See Also:

NetworkClassifierImp.getDefaultConfiguration()

configure

public void configure(Configuration config)

Configure this classifier object. All but the distance function is taken care of by the super class. The distance function is gotten by using a Factory class on the distance.properties file, so the name of the distance function should appear in that file.

Specified by:

configure in interface Configurable

Overrides:

configure in class NetworkClassifierImp

Parameters:

config - The Configuration object used to configure this classifier.

See Also:

NetworkClassifierImp.configure(netkit.util.Configuration)

getShortName

public java.lang.String getShortName()

Returns:

'classDistribRN'

getName

public java.lang.String getName()

Returns:

'Class-Distributional Relational Neighbor NetworkClassifier (classDistribRN)'

getDescription

public java.lang.String getDescription()

Returns:

'No description yet'

induceModel

public void induceModel(Graph graph,
                        DataSplit split)

Induce the cdRN model by finding the 'prototypical' neighborhood for each class of nodes. By default, this neighborhood consists only of the neighboring class labels, but this can easily include any intrinsic or aggregate attributes.

Specified by:

induceModel in interface Classifier

Overrides:

induceModel in class NetworkClassifierImp

Parameters:

graph - Graph whose nodes are to be estimated

split - The split between training and test. Used to get the nodetype and class attribute.

See Also:

NetworkClassifierImp.induceModel(netkit.graph.Graph, netkit.classifiers.DataSplit)

doEstimate

public boolean doEstimate(Node node,
                          double[] result)

Estimate how near this node's neighborhood is to each of the class vectors using a user-specified distance function (cosine by default) and normalize to produce a pseudo distribution.

Specified by:

doEstimate in class NetworkClassifierImp

Parameters:

node - The node to estimate class probabilities for

result - the double array containing the probability estimates that the node belongs to each of the possible class labels.

Returns:

true

See Also:

NetworkClassifierImp.makeVector(netkit.graph.Node, double[])

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object