Package net.sf.tweety.logics.mln.reasoner

Class ApproximateNaiveMlnReasoner

java.lang.Object

net.sf.tweety.logics.mln.reasoner.AbstractMlnReasoner

net.sf.tweety.logics.mln.reasoner.ApproximateNaiveMlnReasoner

All Implemented Interfaces:

QuantitativeReasoner<MarkovLogicNetwork,FolFormula>, Reasoner<java.lang.Double,MarkovLogicNetwork,FolFormula>

public class ApproximateNaiveMlnReasoner
extends AbstractMlnReasoner

This reasoner performs approximate reasoning with MLNs by considering only a subset of all Herbrand interpretations. This subset is chosen by first randomly selecting a set of Herbrand interpretations and then selecting the subset of this set with maximum weights.

Author:

Matthias Thimm

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
private class	ApproximateNaiveMlnReasoner.WeightedHerbrandInterpretation	A Herbrand interpretation with an annotated weight.

Field Summary

Fields

Modifier and Type	Field	Description
private long	maxNumberOfInterpretationsForModel	The maximum number of interpretations used for the model.
private long	maxNumberOfSelectedInterpretations	The maximum number of interpretations selected from the whole set of interpretations.

Constructor Summary

Constructors

Constructor	Description
ApproximateNaiveMlnReasoner(long maxNumberOfSelectedInterpretations, long maxNumberOfInterpretationsForModel)	Creates a new ApproximateNaiveMlnReasoner.

Method Summary

Modifier and Type	Method	Description
java.util.Map<HerbrandInterpretation,java.lang.Double>	computeModel(MarkovLogicNetwork mln, FolSignature signature)	Computes the model of the given MLN wrt.
protected double	doQuery(MarkovLogicNetwork mln, FolFormula query, FolSignature signature)	Performs the actual querying.

Methods inherited from class net.sf.tweety.logics.mln.reasoner.AbstractMlnReasoner

computeWeight, numberOfGroundSatisfactions, query, query

Methods inherited from class java.lang.Object

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

Field Detail

maxNumberOfSelectedInterpretations

private long maxNumberOfSelectedInterpretations

The maximum number of interpretations selected from the whole set of interpretations. Is -1 if all interpretations are to be selected.

maxNumberOfInterpretationsForModel

private long maxNumberOfInterpretationsForModel

The maximum number of interpretations used for the model. Those interpretations are the subset of the interpretations selected with maximum weight. Is -1 if all interpretations are used for the model. It has to be maxNumberOfSelectedInterpretations >= maxNumberOfInterpretationsForModel.

Constructor Detail

ApproximateNaiveMlnReasoner

public ApproximateNaiveMlnReasoner(long maxNumberOfSelectedInterpretations,
                                   long maxNumberOfInterpretationsForModel)

Creates a new ApproximateNaiveMlnReasoner.

Parameters:

maxNumberOfSelectedInterpretations - the maximum number of interpretations selected from the whole set of interpretations. Is -1 if all interpretations are to be selected.

maxNumberOfInterpretationsForModel - the maximum number of interpretations used for the model. Those interpretations are the subset of the interpretations selected with maximum weight. Is -1 if all interpretations are used for the model. It has to be maxNumberOfSelectedInterpretations >= maxNumberOfInterpretationsForModel.

Method Detail

doQuery

protected double doQuery(MarkovLogicNetwork mln,
                         FolFormula query,
                         FolSignature signature)

Description copied from class: AbstractMlnReasoner

Performs the actual querying.

Specified by:

doQuery in class AbstractMlnReasoner

Parameters:

mln - an MLN

query - a fol formula guaranteed to be ground.

signature - the signature

Returns:

the answer of the query.

computeModel

public java.util.Map<HerbrandInterpretation,java.lang.Double> computeModel(MarkovLogicNetwork mln,
                                                                                 FolSignature signature)

Computes the model of the given MLN wrt. the optimization parameters

Parameters:

mln - some mln

signature - some signature

Returns:

the model of the given MLN wrt. the optimization parameters.