nars.inference
Class BudgetFunctions

java.lang.Object
  nars.inference.UtilityFunctions
      nars.inference.BudgetFunctions

public final class BudgetFunctions
extends UtilityFunctions

Budget functions for resources allocation

Constructor Summary

BudgetFunctions()

Method Summary

static void	activate(Concept concept, BudgetValue budget) Activate a concept by an incoming TaskLink
static BudgetValue	backward(TruthValue truth) Backward inference result and adjustment, stronger case
static BudgetValue	backwardWeak(TruthValue truth) Backward inference result and adjustment, weaker case
private static BudgetValue	budgetInference(float qual, int complexity) Common processing for all inference step
static BudgetValue	compoundBackward(Term content) Backward inference with CompoundTerm conclusion, stronger case
static BudgetValue	compoundBackwardWeak(Term content) Backward inference with CompoundTerm conclusion, weaker case
static BudgetValue	compoundForward(TruthValue truth, Term content) Forward inference with CompoundTerm conclusion
static BudgetValue	distributeAmongLinks(BudgetValue b, int n) Distribute the budget of a task among the links to it
static void	forget(BudgetValue budget, float forgetRate, float relativeThreshold) Decrease Priority after an item is used, called in Bag
(package private) static BudgetValue	forward(TruthValue truth) Forward inference result and adjustment
static void	merge(BudgetValue baseValue, BudgetValue adjustValue) Merge an item into another one in a bag, when the two are identical except in budget values
static float	rankBelief(Judgment judg) Determine the rank of a judgment by its confidence and originality (stamp length)
(package private) static BudgetValue	revise(TruthValue tTruth, TruthValue bTruth, TruthValue truth, Task task, boolean feedbackToLinks) Evaluate the quality of a revision, then de-prioritize the premises
(package private) static BudgetValue	solutionEval(Sentence problem, Judgment solution, Task task) Evaluate the quality of a belief as a solution to a problem, then reward the belief and de-prioritize the problem
static BudgetValue	temporalIndCom(BudgetValue b1, BudgetValue b2, TruthValue t) Special treatment for temporal induction and comparison among recent event, no feedback
static float	truthToQuality(TruthValue t) Determine the quality of a judgment by its truth value alone
(package private) static BudgetValue	update(Task task, TruthValue bTruth) Update a belief

Methods inherited from class nars.inference.UtilityFunctions

and, aveAri, aveGeo, or, w2c

Methods inherited from class java.lang.Object

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

Constructor Detail

BudgetFunctions

public BudgetFunctions()

Method Detail

truthToQuality

public static float truthToQuality(TruthValue t)

Determine the quality of a judgment by its truth value alone

Mainly decided by confidence, though binary judgment is also preferred

Parameters:

t - The truth value of a judgment

Returns:

The quality of the judgment, according to truth value only

rankBelief

public static float rankBelief(Judgment judg)

Determine the rank of a judgment by its confidence and originality (stamp length)

Parameters:

judg - The judgment to be ranked

Returns:

The rank of the judgment, according to truth value only

solutionEval

static BudgetValue solutionEval(Sentence problem,
                                Judgment solution,
                                Task task)

Evaluate the quality of a belief as a solution to a problem, then reward the belief and de-prioritize the problem

Parameters:

problem - The problem (question or goal) to be solved

solution - The belief as solution

task - The task to be immediatedly processed, or null for continued process

Returns:

The budget for the new task which is the belief activated, if necessary

revise

static BudgetValue revise(TruthValue tTruth,
                          TruthValue bTruth,
                          TruthValue truth,
                          Task task,
                          boolean feedbackToLinks)

Evaluate the quality of a revision, then de-prioritize the premises

Parameters:

tTruth - The truth value of the judgment in the task

bTruth - The truth value of the belief

truth - The truth value of the conclusion of revision

task - The task to be immediatedly or continuely processed

Returns:

The budget for the new task

update

static BudgetValue update(Task task,
                          TruthValue bTruth)

Update a belief

Parameters:

task - The task containing new belief

bTruth - Truth value of the previous belief

Returns:

Budget value of the updating task

distributeAmongLinks

public static BudgetValue distributeAmongLinks(BudgetValue b,
                                               int n)

Distribute the budget of a task among the links to it

Parameters:

b - The original budget

n - Number of links

Returns:

Budget value for each link

activate

public static void activate(Concept concept,
                            BudgetValue budget)

Activate a concept by an incoming TaskLink

Parameters:

concept - The concept

budget - The budget for the new item

forget

public static void forget(BudgetValue budget,
                          float forgetRate,
                          float relativeThreshold)

Decrease Priority after an item is used, called in Bag

After a constant time, p should become d*p. Since in this period, the item is accessed c*p times, each time p-q should multiple d^(1/(c*p)). The intuitive meaning of the parameter "forgetRate" is: after this number of times of access, priority 1 will become d, it is a system parameter adjustable in run time.

Parameters:

budget - The previous budget value

forgetRate - The budget for the new item

relativeThreshold - The relative threshold of the bag

merge

public static void merge(BudgetValue baseValue,
                         BudgetValue adjustValue)

Merge an item into another one in a bag, when the two are identical except in budget values

Parameters:

baseValue - The budget value to be modified

adjustValue - The budget doing the adjusting

forward

static BudgetValue forward(TruthValue truth)

Forward inference result and adjustment

Parameters:

truth - The truth value of the conclusion

Returns:

The budget value of the conclusion

backward

public static BudgetValue backward(TruthValue truth)

Backward inference result and adjustment, stronger case

Parameters:

truth - The truth value of the belief deriving the conclusion

Returns:

The budget value of the conclusion

backwardWeak

public static BudgetValue backwardWeak(TruthValue truth)

Backward inference result and adjustment, weaker case

Parameters:

truth - The truth value of the belief deriving the conclusion

Returns:

The budget value of the conclusion

compoundForward

public static BudgetValue compoundForward(TruthValue truth,
                                          Term content)

Forward inference with CompoundTerm conclusion

Parameters:

truth - The truth value of the conclusion

content - The content of the conclusion

Returns:

The budget of the conclusion

compoundBackward

public static BudgetValue compoundBackward(Term content)

Backward inference with CompoundTerm conclusion, stronger case

Parameters:

content - The content of the conclusion

Returns:

The budget of the conclusion

compoundBackwardWeak

public static BudgetValue compoundBackwardWeak(Term content)

Backward inference with CompoundTerm conclusion, weaker case

Parameters:

content - The content of the conclusion

Returns:

The budget of the conclusion

budgetInference

private static BudgetValue budgetInference(float qual,
                                           int complexity)

Common processing for all inference step

Parameters:

qual - Quality of the inference

complexity - Syntactic complexity of the conclusion

Returns:

Budget of the conclusion task

temporalIndCom

public static BudgetValue temporalIndCom(BudgetValue b1,
                                         BudgetValue b2,
                                         TruthValue t)

Special treatment for temporal induction and comparison among recent event, no feedback

Parameters:

b1 - Budget value of the first premise

b2 - Budget value of the second premise

t - Truth value of the conclusion

Returns:

Budget value of the conclusion task