netkit.classifiers.relational
Class WeightedVoteRelationalNeighbor

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

All Implemented Interfaces:

Classifier, NetworkClassifier, Configurable

public final class WeightedVoteRelationalNeighbor
extends NetworkClassifierImp

weighted-vote Relational Neighbor Classifier (wvRN). Reference:

• Sofus A. Macskassy, Foster Provost (2007). Classification in Networked Data: A toolkit and a univariate case study. Journal of Machine Learning, 8(May):935-983, 2007.

NOTE: This builds a model only on the neighborhood class labels. Parameters:

.laplace=None
.laplaceone=false
.lfactor=1.0
.rfactor=0.0

Equivalent hard-coded parameters:

.aggregation=ClassOnly
.aggregators=count

Author:

Sofus A. Macskassy

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

WeightedVoteRelationalNeighbor()

Method Summary

void	configure(Configuration conf) Configures the classifier with respect to laplace correction: whether to have it (and what kind) and whether tu use it only on the first iteration of collective inferencing.
boolean	doEstimate(Node node, double[] estimation) This is the final estimation method that will be called and the only estimation method that sub-classes should implement.
Configuration	getDefaultConfiguration() Creates and returns a default configuration, which only includes the laplaceonce, laplace and lfactor properties (the only ones used in this classifier as nothing else is not configurable).
java.lang.String	getDescription()
java.lang.String	getName()
java.lang.String	getShortName()
void	induceModel(Graph graph, DataSplit split) wvRN has no model, so this only initializes what needs to be done for laplace correction (in addition to whatever the superclass does).
void	initializeRun(Estimate currPrior, Node[] unknowns) This initializes wvRN for the next collective inference iteration by setting up the laplace correction, if needed, for the current iteration.
java.lang.String	toString()

Methods inherited from class netkit.classifiers.relational.NetworkClassifierImp

classify, estimate, estimate, estimate, estimate, generateAggregators, getAttributeNames, includeClassAttribute, 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

WeightedVoteRelationalNeighbor

public WeightedVoteRelationalNeighbor()

Method Detail

getShortName

public java.lang.String getShortName()

Returns:

return "wvRN"

getName

public java.lang.String getName()

Returns:

return "Weighted Vote Relational Neighbor NetworkClassifier (wvRN)"

getDescription

public java.lang.String getDescription()

Returns:

return "See reference: [Macskassy, Provost] 'Classification in Networked Data'"

getDefaultConfiguration

public Configuration getDefaultConfiguration()

Creates and returns a default configuration, which only includes the laplaceonce, laplace and lfactor properties (the only ones used in this classifier as nothing else is not configurable). By default, there is no laplace correction.

Specified by:

getDefaultConfiguration in interface Configurable

Overrides:

getDefaultConfiguration in class NetworkClassifierImp

Returns:

configation object

configure

public void configure(Configuration conf)

Configures the classifier with respect to laplace correction: whether to have it (and what kind) and whether tu use it only on the first iteration of collective inferencing.

Specified by:

configure in interface Configurable

Overrides:

configure in class NetworkClassifierImp

Parameters:

conf - Configuration object to use to configure the classifier

See Also:

NetworkClassifierImp.aggregation

initializeRun

public void initializeRun(Estimate currPrior,
                          Node[] unknowns)

This initializes wvRN for the next collective inference iteration by setting up the laplace correction, if needed, for the current iteration. If the laplace correction is external, then it caches the class priors on the first iteration as they are the external priors. If we should only use laplace correction the first time, then laplace correction is set to None after the first iteration

Specified by:

initializeRun in interface NetworkClassifier

Overrides:

initializeRun in class NetworkClassifierImp

Parameters:

currPrior - The current priors for all nodes in the graph

unknowns - The list of nodes whose labels are unknown

induceModel

public void induceModel(Graph graph,
                        DataSplit split)

wvRN has no model, so this only initializes what needs to be done for laplace correction (in addition to whatever the superclass does).

Specified by:

induceModel in interface Classifier

Overrides:

induceModel in class NetworkClassifierImp

Parameters:

graph - The graph to induce a model over

split - The data split identifying which nodes have known and unknown labels

See Also:

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

doEstimate

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

Description copied from class: NetworkClassifierImp

This is the final estimation method that will be called and the only estimation method that sub-classes should implement.

Specified by:

doEstimate in class NetworkClassifierImp

Parameters:

node -

estimation -

Returns:

true

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object