com.ibm.able.rules
Class AbleFuzzyInferenceEngine

java.lang.Object
  com.ibm.able.rules.AbleInferenceEngine
      com.ibm.able.rules.AbleFuzzyInferenceEngine

All Implemented Interfaces:

java.io.Serializable

public class AbleFuzzyInferenceEngine

extends AbleInferenceEngine

implements java.io.Serializable

This class provides the control code and inferencing algorithm for processing a set of conditional rules in a ruleblock containing Fuzzy linguistic variables and hedges using data-driven or forward chaining and fuzzy inferencing logic.

The AbleFuzzyInferenceEngine processes AbleAssertionRules and AbleConditionalRules and ignores all others.

The control parameter ARL.InferenceMethod can be used to specify several alternative fuzzy inference algorithms. These include:

1. FUZZY_ADD The default inference method.
2. MIN_MAX
3. PRODUCT_OR

*

The control parameter ARL.CorrelationMethod can be used to specify alternative correlation algorithms. These include:

1. Product The default correlation method.
2. Minimum

The control parameter ARL.DefuzzifyMethod can be used to specify several alternative defuzzification algorithms. These include:

1. Centriod The default defuzzification method.
2. MaxHeight

The major processing method is infer() which is called by the AbleRuleSet bean as part of the normal bean process() method.

Instances of this class are not directly constructed by the user but are dynamically created by the AbleRuleBlock init() method.

See Also:

AbleRuleSet, AbleRuleBlock, AbleAssertionRule, AbleConditionalRule, Serialized Form

Field Summary

protected double	myAlphaCut Ruleset alphaCut is the default alphaCut for fuzzy sets.
protected int	myCorrelationMethod Specifies the manner in which a rule's consequent is correlated with the truth values of the rule's antecedents.
protected int	myDefuzzifyMethod Specifies the method used to turn a fuzzy value into a crisp number.
protected int	myFuzzyInferenceMethod Specifies the inference (implication) used to process fuzzy rules Value is a DefuzzifyMethod constant.

Fields inherited from class com.ibm.able.rules.AbleInferenceEngine

myDebugLevel, myRuleBlock, myRuleSet, myRulesFiredCount, myTracer, myWorkingMemory

Constructor Summary

AbleFuzzyInferenceEngine(AbleRuleSet theRuleSet, AbleLogger theTracer, AbleRuleBlock theRuleBlock)
Create a new inference engine.

Method Summary

protected void	evalAssertionClause(AbleAssertionClause theClause) Evaluate the specified assertion clause.
protected void	evaluateDoActions(AbleRule theRule, java.util.BitSet theFactBase, double theTruthValueMin)
double	getAlphaCut() Retrieve the current alphacut threshold.
java.lang.Object	getControlParameter(java.lang.String theControlParameter) Returns the specified control parameter on the associated inference engine.
int	getCorrelationMethod() Retrieve the current correlation method.
int	getDefuzzifyMethod() Retrieve the current defuzzification method.
int	getFuzzyInferenceMethod() Retrieve the current fuzzy inference method.
void	infer(AbleRuleBlock theRuleBlock) Fire the rules in the ruleblock using fuzzy inferencing.
boolean	isRuleBlockValid(AbleRuleBlock theRuleBlock) Make sure the current ruleblock is "executable" by this inference engine.
void	processRuleBlock(AbleRuleBlock theRuleBlock) Fire all rules in the specified rule block in sequential order.
void	setAlphaCut(double theAlphaCut) Set the alphacut threshold to the specified value.
void	setControlParameter(java.lang.String theControlParameter, java.lang.Object theValue) Set the specified control parameter on the associated inference engine.
void	setCorrelationMethod(int theCorrelationMethod) Set the correlation method to the specified type.
void	setDefuzzifyMethod(int theDefuzzifyMethod) Set the defuzzification method to the specified type.
void	setFuzzyInferenceMethod(int theInferenceMethod) Set the inference method to the specified type.

Methods inherited from class com.ibm.able.rules.AbleInferenceEngine

allocateWorkingMemory, askUser, conditionalRulesValid, evalAntecedentClause, evalAntecedentExpression, evalAssertionRule, evalConditionalRule, evalConsequentClause, evalDoUntilRule, evalDoWhileRule, evalForLoopRule, evalIfThenElseRule, evalPatternMatchRule, evalSwitchRule, evaluateDoActions, evalWhileDoRule, fireEngineBreakpoint, firePropertyChange, getRuleSet, getRulesFiredCount, getWorkingMemory, init, processAssertions, processConsequent, processConsequent, processConsequentClauses, reset, resetRuleSet, setControlParameters, setDebugLevel, setInferenceContext, setInferenceContext, setInferenceContext, setRuleSet, setWorkingMemory, startDebugConsole, stopDebugConsole, toString

Methods inherited from class java.lang.Object

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

Field Detail

myFuzzyInferenceMethod

protected int myFuzzyInferenceMethod

Specifies the inference (implication) used to process fuzzy rules Value is a DefuzzifyMethod constant. Initially set to FuzzyInferenceMethodDefault.

myAlphaCut

protected double myAlphaCut

Ruleset alphaCut is the default alphaCut for fuzzy sets. Specifies the floor beneath which truth values are considered to be zero. Initially set to AlphaCutDefault.

myCorrelationMethod

protected int myCorrelationMethod

Specifies the manner in which a rule's consequent is correlated with the truth values of the rule's antecedents. Value is a CorrelationMethod constant. Initially set to CorrelationMethodDefault.

myDefuzzifyMethod

protected int myDefuzzifyMethod

Specifies the method used to turn a fuzzy value into a crisp number. Value is a DefuzzifyMethod constant. Initially set to DefuzzifyMethodDefault.

Constructor Detail

AbleFuzzyInferenceEngine

public AbleFuzzyInferenceEngine(AbleRuleSet theRuleSet,
                                AbleLogger theTracer,
                                AbleRuleBlock theRuleBlock)

Create a new inference engine.

Parameters:

theRuleSet - A ruleset whose context is used for inferencing by this inference engine.

theTracer - A tracer object to which this inference engine can log information. This may be null, in which case no information is logged.

theRuleBlock - A ruleblock whose rules are to be used for inferencing by this inference engine.

Method Detail

isRuleBlockValid

public boolean isRuleBlockValid(AbleRuleBlock theRuleBlock)

Make sure the current ruleblock is "executable" by this inference engine.

The ruleblock is invalid if any of the following are present in any ruleblock:

If-then-else rules

Pattern match rules

Predicate rules

Iteration rules

All reasons for validation failures are appended to the AbleRuleSet's myIsExReason variable. If validation fails, the inference engine must call myRuleSet.addIsExReason(java.lang.String) to append the reason. Reasons are accumulatve.

Specified by:

isRuleBlockValid in class AbleInferenceEngine

Returns:

true, if the ruleset is OK; false, if the ruleset is not executable.

infer

public void infer(AbleRuleBlock theRuleBlock)
           throws AbleDataException

Fire the rules in the ruleblock using fuzzy inferencing.

The infer() method is invoked by the AbleRuleSet process() code.

The order of processing is as follows:

1. Evaluate all assertion rules sequentially.
2. Evaluate all conditional rules using fuzzy inferencing.

Specified by:

infer in class AbleInferenceEngine

Parameters:

theRuleBlock - The ruleblock object to process.

Throws:

AbleDataException - When any error occurs.

processRuleBlock

public void processRuleBlock(AbleRuleBlock theRuleBlock)
                      throws AbleDataException

Fire all rules in the specified rule block in sequential order.

This method is called from a ruleset's invokeRuleBlock() method, and that method is called when a rule that uses the built-in invokeRuleBlock("name") call literal is evaluated.

Overrides:

processRuleBlock in class AbleInferenceEngine

Parameters:

theRuleBlock - The rule block from which rules are to be processed.

Throws:

AbleDataException - when the data types are incompatible, or when the relational operator is invalid for the current situation or unrecognized.

evaluateDoActions

protected void evaluateDoActions(AbleRule theRule,
                                 java.util.BitSet theFactBase,
                                 double theTruthValueMin)
                          throws AbleDataException

Throws:

AbleDataException

evalAssertionClause

protected void evalAssertionClause(AbleAssertionClause theClause)
                            throws AbleDataException

Evaluate the specified assertion clause.

Overrides:

evalAssertionClause in class AbleInferenceEngine

Parameters:

theClause - an assertion clause to evaluate.

Throws:

AbleDataException - when the clause's data objects are incompatible, or when the relational operator is invalid for the current situation or unrecognized.

setControlParameter

public void setControlParameter(java.lang.String theControlParameter,
                                java.lang.Object theValue)
                         throws AbleDataException

Set the specified control parameter on the associated inference engine.

Specified by:

setControlParameter in class AbleInferenceEngine

Throws:

AbleDataException

getControlParameter

public java.lang.Object getControlParameter(java.lang.String theControlParameter)
                                     throws AbleDataException

Returns the specified control parameter on the associated inference engine.

Specified by:

getControlParameter in class AbleInferenceEngine

Throws:

AbleDataException

getAlphaCut

public final double getAlphaCut()

Retrieve the current alphacut threshold.

Returns:

The current alphacut.

setAlphaCut

public void setAlphaCut(double theAlphaCut)
                 throws AbleDataException

Set the alphacut threshold to the specified value.

Parameters:

theAlphaCut - The new alphacut threshold.

Throws:

AbleDataException - If the specified alphacut value is not between 0.0 (exclusive) and 1.0 (exclusive).

getCorrelationMethod

public final int getCorrelationMethod()

Retrieve the current correlation method.

Returns:

The current correlation method.

setCorrelationMethod

public void setCorrelationMethod(int theCorrelationMethod)
                          throws AbleDataException

Set the correlation method to the specified type.

Parameters:

theCorrelationMethod - The new correlation method.

Throws:

AbleDataException - If the specified correlation method is not recognised.

getDefuzzifyMethod

public final int getDefuzzifyMethod()

Retrieve the current defuzzification method.

Returns:

The current defuzzification method.

setDefuzzifyMethod

public void setDefuzzifyMethod(int theDefuzzifyMethod)
                        throws AbleDataException

Set the defuzzification method to the specified type.

Parameters:

theDefuzzifyMethod - The new defuzzification method.

Throws:

AbleDataException - If the specified defuzzification method is not recognised.

setFuzzyInferenceMethod

public void setFuzzyInferenceMethod(int theInferenceMethod)
                             throws AbleDataException

Set the inference method to the specified type.

Parameters:

theInferenceMethod - The new fuzzy inference method. Fuzzy Inference methods are enumerated in this class.

Throws:

AbleDataException - If the specified fuzzy inference method is not recognised.

See Also:

getFuzzyInferenceMethod()

getFuzzyInferenceMethod

public final int getFuzzyInferenceMethod()

Retrieve the current fuzzy inference method.

Returns:

The current fuzzy inference method.

See Also:

setFuzzyInferenceMethod(int)