Package net.sf.tweety.logics.rcl.reasoner

Class SimpleRelationalCReasoner

java.lang.Object

net.sf.tweety.logics.rcl.reasoner.SimpleRelationalCReasoner

All Implemented Interfaces:

ModelProvider<RelationalConditional,RclBeliefSet,RelationalRankingFunction>, QualitativeReasoner<RclBeliefSet,FolFormula>, Reasoner<java.lang.Boolean,RclBeliefSet,FolFormula>

public class SimpleRelationalCReasoner
extends java.lang.Object
implements QualitativeReasoner<RclBeliefSet,FolFormula>, ModelProvider<RelationalConditional,RclBeliefSet,RelationalRankingFunction>

This class models a relational brute force c-reasoner for relational conditional logic. Reasoning is performed by computing a minimal c-representation for the given knowledge base.
A c-representation for a conditional knowledge base R={r1,...,rn} is a ranking function k such that k accepts every conditional in R (k |= R) and if there are numbers k0,k1+,k1-,...,kn+,kn- with
k(w)=k0 + \sum_{w verifies ri} ki+ + \sum_{w falsifies ri} kj- for every w. A c-representation is minimal if k0+...+kn- is minimal.
The c-representation is computed using a brute force approach.

See Gabriele Kern-Isberner. Conditionals in nonmonotonic reasoning and belief revision. Lecture Notes in Computer Science, Volume 2087. 2001.

See also [Kern-Isberner,Thimm, "A Ranking Semantics for Relational Defaults", in preparation].

Author:

Matthias Thimm

Field Summary

Fields

Modifier and Type	Field	Description
private static org.slf4j.Logger	log	Logger.
private boolean	simple	indicates whether the computed c-representation is simple.

Constructor Summary

Constructors

Constructor	Description
SimpleRelationalCReasoner(boolean simple)	Creates a new relational c-representation reasoner.

Method Summary

Modifier and Type	Method	Description
private RelationalRankingFunction	constructRankingFunction(java.lang.Integer[] kappa, java.util.Map<java.lang.Integer,RelationalConditional> indexToConditional, FolSignature signature)	Constructs a ranking function with the given kappa values [k1+,k1-,...,kn+,kn-], i.e.
RelationalRankingFunction	getModel(RclBeliefSet bbase)	Returns a single (dedicated) model of the given belief base.
RelationalRankingFunction	getModel(RclBeliefSet bs, FolSignature signature)	Retrieves the C representation of the given belief set wrt.
java.util.Collection<RelationalRankingFunction>	getModels(RclBeliefSet bbase)	Returns a characterizing model of the given belief base
private java.lang.Integer[]	increment(java.util.List<java.lang.Integer[]> kappa_all, int numConditionals)	This method increments the given array by one value.
private boolean	incrementStep(java.lang.Integer[] kappaRow)	This method increments the given vector (which composes of exactly one "1" entry and zeros otherwise), e.g.
java.lang.Boolean	query(RclBeliefSet bs, FolFormula query)	Queries the given belief base for the given formula.
java.lang.Boolean	query(RclBeliefSet bs, FolFormula query, FolSignature signature)	Queries the given belief set wrt.

Methods inherited from class java.lang.Object

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

Field Detail

log

private static org.slf4j.Logger log

Logger.

simple

private boolean simple

indicates whether the computed c-representation is simple.

Constructor Detail

SimpleRelationalCReasoner

public SimpleRelationalCReasoner(boolean simple)

Creates a new relational c-representation reasoner.

Parameters:

simple - whether the computed c-representation is simple.

Method Detail

query

public java.lang.Boolean query(RclBeliefSet bs,
                               FolFormula query,
                               FolSignature signature)

Queries the given belief set wrt. the given signature.

Parameters:

bs - some belief set

query - some query

signature - some signature

Returns:

true iff the query is true

query

public java.lang.Boolean query(RclBeliefSet bs,
                               FolFormula query)

Description copied from interface: Reasoner

Queries the given belief base for the given formula.

Specified by:

query in interface QualitativeReasoner<RclBeliefSet,FolFormula>

Specified by:

query in interface Reasoner<java.lang.Boolean,RclBeliefSet,FolFormula>

Parameters:

bs - a belief base

query - a formula

Returns:

the answer to the query

constructRankingFunction

private RelationalRankingFunction constructRankingFunction(java.lang.Integer[] kappa,
                                                           java.util.Map<java.lang.Integer,RelationalConditional> indexToConditional,
                                                           FolSignature signature)

Constructs a ranking function with the given kappa values [k1+,k1-,...,kn+,kn-], i.e. for every possible world w set
k(w)=k0 + \sum_{ w verifies ri} ki+ + \sum_{w falsifies ri} kj-

Parameters:

kappa - the kappa values

indexToConditional - a map mapping indices to conditionals

signature - the signature

Returns:

the ranking function

increment

private java.lang.Integer[] increment(java.util.List<java.lang.Integer[]> kappa_all,
                                      int numConditionals)

This method increments the given array by one value.

Parameters:

kappa_all - the values of all kappas

numConditionals - the number of conditionals

Returns:

an array of integers.

incrementStep

private boolean incrementStep(java.lang.Integer[] kappaRow)

This method increments the given vector (which composes of exactly one "1" entry and zeros otherwise), e.g. [0,0,1,0] -> [0,0,0,1] and [0,0,0,1] -> [1,0,0,0]

Parameters:

kappaRow - a vector of zeros and one "1"

Returns:

"true" if there is an overflow, i.e. when [0,0,0,1] -> [1,0,0,0], otherwise "false".

getModels

public java.util.Collection<RelationalRankingFunction> getModels(RclBeliefSet bbase)

Description copied from interface: ModelProvider

Returns a characterizing model of the given belief base

Specified by:

getModels in interface ModelProvider<RelationalConditional,RclBeliefSet,RelationalRankingFunction>

Parameters:

bbase - some belief base

Returns:

the (selected) models of the belief base

getModel

public RelationalRankingFunction getModel(RclBeliefSet bbase)

Description copied from interface: ModelProvider

Returns a single (dedicated) model of the given belief base. If the implemented method allows for more than one dedicated model, the selection may be non-deterministic.

Specified by:

getModel in interface ModelProvider<RelationalConditional,RclBeliefSet,RelationalRankingFunction>

Parameters:

bbase - some belief base

Returns:

a selected model of the belief base.

getModel

public RelationalRankingFunction getModel(RclBeliefSet bs,
                                          FolSignature signature)

Retrieves the C representation of the given belief set wrt. the given signature

Parameters:

bs - some belief set

signature - some signature

Returns:

the c representation (a ranking function)