Assessing Certainty Factors in Decision-Making !!

Abstract:

A Certainty Factor (CF) is a quantitative measurement of how strongly the antecedent of a rule supports its conclusion. CFs are used in rule-based systems to represent uncertain knowledge and assign weight to facts or pieces of knowledge. 

 

CF value

Meaning

-1

Certainly false

+1

Definitely true

Intermediate values

Varying degrees of certainty

0

Unknown

CFs were first introduced in the MYCIN expert system for medical diagnosis. The developers of MYCIN abandoned Bayes' Theorem and the p-function because they felt that experts' knowledge and intuition defied rigorous analysis. 

 

The CF model has theoretical and practical limitations. A belief network representation is similar to the CF model but is grounded in probability theory. The belief network representation has several advantages over the CF model, including overcoming many of the limitations of the CF model. 

 

Keywords 

Certainty Factors, Certainly False, Definitely True, CF Model, Knowledge and Intuition

Learning Outcomes

After undergoing this article you will be able to understand the following:

1. What is Certainty Factor (CF)?

2. Why Certainty Factor (CF) is important?

3. How Certainty Factor (CF) works?

4. What's the characteristics of Certainty Factor (CF)?

5. How many types of Certainty Factor (CF) are there?

6. What is the components of 

Certainty Factor (CF)?

7. Methods of Certainty Factor (CF)

8. Techniques of Certainty Factor (CF)

9. Applications of Certainty Factor (CF)

10. How Certainty Factor (CF) benefits an organisation?

11. What are the limitations of Certainty Factor (CF)?

12. Conclusions

13. FAQs

References

1. What is Certainty Factor (CF)?

In artificial intelligence (AI), a certainty factor (CF) is a numerical value that represents how likely a statement or event is to occur. CFs are used to represent uncertain knowledge in expert systems. 

 

CFs are similar to probability, but there are some differences:

Humans: Humans can determine if a proposition is true or false based on how certain they are about their observations.

Machines: Machines lack this analytical capability. 

 

CFs are typically represented as a range of negative to positive values, such as -100 to 100: 

 

-100: Represents a complete lack of belief in something 

 

100: Represents an absolute belief in a rule or value 

 

Around 0: Indicates that there is little evidence either for or against the hypothesis 

 

CFs were first introduced in the MYCIN expert system for medical diagnosis. 

 

2. Why Certainty Factor (CF) is important?

Certainty factor (CF) is important in artificial intelligence (AI) because it's a way to measure the likelihood of a hypothesis being true based on evidence: 

 

Representing uncertainty

CF is a theory that uses numeric values between -1 and 1 to represent the certainty of statements or hypotheses. 

 

Efficient inference

CF allows for efficient inference in uncertain situations by combining certainty factors from multiple rules. 

 

Expert systems

CF is often used in traditional expert systems (TES) in the medical industry to compute symptoms and identify inference solutions. 

 

CF is calculated based on: Measure of belief, Measure of disbelief, and Formulas to combine certainty factors from multiple pieces of evidence. 

 

However, CF has some limitations, such as difficulty accurately assigning certainty values and a limited numeric range. The belief network representation is a similar model to CF, but it's grounded in probability theory and has several advantages over CF. 

3. How Certainty Factor (CF) works ?

The Certainty Factor (CF) algorithm is a method used in artificial intelligence (AI) to manage uncertainty in expert systems: 

 

How it works

The CF algorithm measures the strength of evidence supporting a conclusion by considering factors like the uncertainty or confidence of the input data and the rules used. 

 

What it does

The CF algorithm helps expert systems calculate the confidence or certainty of conclusions based on available information. It also allows for efficient inference in uncertain situations by combining certainty factors from multiple rules. 

 

How it's measured

The CF is a numeric value that lies between -1 and 1. A CF of -1 means the statement is never true, and a CF of +1 means the statement is always true. 

 

Limitations

The belief network model is similar to the CF model but is grounded in probability theory. The belief network model overcomes many of the limitations of the CF model. 

 

4. What's the characteristics of Certainty Factor (CF)?

Here are some features of certainty factors in artificial intelligence (AI): 

 

Numerical value: A certainty factor (CF) is a numeric value that indicates how likely a statement or event is to occur. 

 

Range: CFs range from -1.0 to +1.0. 

 

Meaning: A CF of -1.0 means the statement is never true, while a CF of +1.0 means the statement is always true. A CF of 0 means the agent is unaware of the condition or occurrence. 

 

Based on evidence: CFs are based on evidence or an expert's assessment. 

 

Inference: CFs allow for efficient inference in uncertain situations by combining CFs from multiple rules. 

 

Threshold value: The minimal certainty factor, also known as the threshold value, is used to determine whether an assertion is true or untrue. For example, if the threshold value is 0.4 and the CF is less than that, the assertion is incorrect. 

 

5. How many types of Certainty Factor (CF) are there?

Here are some types of certainty factors (CF) in artificial intelligence (AI):    Single premise rules The certainty factor for rules with a single premise is calculated as \(CF(H,E)=CF(E)*CF(rule)=CF(user)*CF(expert)\). Multiple premise rules The certainty factor for rules with multiple premises is calculated as: \(CF(AANDB)=Minimum(CF(a),CF(b))*CF(rule)\) \(CF(AORB)=Maximum(CF(a),CF(b))*CF(rule)\)  IOPsciencehttps://iopscience.iop.orgImplementation of the Certainty Factor Method for Early Detection of ...The certainty Factor for rules with a single premise (single premise rules): CF (H, E) = CF (E) * CF (rule) = CF (user) * CF (expert) (2) 2. The certainty Facto...       CFs are used in expert systems to represent uncertain knowledge. They are typically represented as a value between -100 and 100, where -100 indicates a complete lack of belief and 100 indicates absolute belief.  ScienceDirect.comhttps://www.sciencedirect.comCertainty Factor - an overview | ScienceDirect TopicsPurdue Universityhttps://engineering.purdue.eduCertainty Factors - Purdue EngineeringCertainty factors normally range from some negative value to some positive value. For example from -100 to 100. A certainty factor of -100 would represent a com...          CFs have several advantages, including:    They mimic how people manipulate numeric measures of uncertainty. They provide efficient inference in the face of uncertainty.  ScienceDirect.comhttps://www.sciencedirect.comCertainty Factor - an overview | ScienceDirect Topics       CFs were first introduced in the MYCIN expert system for medical diagnosis.  

6. What is the components of 

Certainty Factor (CF)?

The two components of a certainty factor are a measure of belief and a measure of disbelief. Certainty factors are numerical values that represent the likelihood of a statement or hypothesis being true, based on evidence. They are used in artificial intelligence systems to handle uncertainty in rules and user-supplied information. 

 

Here are some properties of certainty factors:

Range

Certainty factors can range from -1 to 1. A value of -100 represents a complete lack of belief, while a value of 100 represents absolute belief.

Axioms

Certainty factors have two fundamental axioms:

A certainty factor is a single real number.

A certainty factor can take on any value between -1 and 1. 

7. Methods of Certainty Factor (CF)

The certainty factor (CF) model is a method for managing uncertainty in rule-based systems. It uses a quantitative measure to show how strongly the antecedent of a rule supports its conclusion. The CF model is used in expert systems to represent uncertain knowledge and calculate the confidence or certainty in conclusions. 

 

Here are some methods of certainty factor: 

 

CF algorithm

This algorithm considers factors that affect the certainty of a statement or decision, such as the level of confidence or uncertainty in the input data or the rules used. 

 

CF(R1,R2) formula

This formula is used to implement a disease diagnosis expert system. The formula is CF(R1,R2) = CF(R1) + CF(R2) – [ (CF(R1) x CF(R2) )]. 

 

Belief network

This representation is similar to the CF model but is grounded in probability theory. It has several advantages over the CF model, including overcoming limitations and providing a promising approach to the practical construction of expert systems. 

 

The CF model was introduced in the MYCIN expert system for medical diagnosis. It is frequently used by traditional expert systems (TES) in the medical industry to compute several symptoms and identify the inference solutions. 

 

8. Techniques of handling Uncertainty to convert into Certainty Factor (CF)

There are many ways to handle uncertainty, including: 

 

Accepting uncertainty: Accept that uncertainty is a normal part of life and focus on what you can control. 

 

Planning: Set short-term goals and prepare for different possible outcomes. 

 

Seeking support: Ask friends, family, or professionals for guidance and reassurance. 

 

Staying informed: Learn from past experiences and stay informed. 

 

Maintaining a routine: Eat regularly and maintain a consistent sleep schedule. 

 

Taking action: Find ways to engage with values that are important to you. For example, if you value family, you could call an aunt you haven't spoken to in a while. 

 

Some mathematical models and tools that can help with uncertainty include: 

 

Fuzzy logic

A mathematical model that can help with inference from abstract and subjective ideas. 

 

Fuzzy sets

A tool that can help express people's hesitations and deal with uncertainty. 

 

Bayesian inference

A general-purpose method for handling uncertainty in experimental datasets. 

 

9. Applications of Certainty Factor (CF)

The certainty factor (CF) algorithm is a technique used in expert systems to manage uncertainty in conclusions or inferences. It's used to measure how strongly evidence supports a conclusion and to calculate the confidence in the resulting conclusions. Here are some applications of the CF algorithm: 

 

Meat classification

In an expert system for classifying meat, the CF algorithm can help with uncertainty in distinguishing between beef and pork. For example, if some physical characteristics are contradictory, the CF algorithm can consider the uncertainty associated with the rules or input data. 

 

Medical diagnosis

In traditional expert systems (TES), the CF algorithm can be used to compute symptoms and identify solutions to infer the likelihood of a particular ailment. 

 

Cat disease diagnosis

A web-based expert system can use the CF algorithm and forward chaining methods to diagnose cat diseases. The system's knowledge base uses 16 types of symptoms and 6 types of diseases. 

 

The CF algorithm is a quantitative measure that's associated with rules in knowledge-based systems. It allows for efficient inference in uncertain situations by combining certainty factors from multiple rules. 

 

10. How Certainty Factor (CF) benefits an organisation?

Certainty factors stated trust in an event based on evidence or judgment from an expert. Certainty factor uses the value to assume an expert's degree of confidence in data. There are concepts of beliefs and uncertainties that are then formulated in the basic formula

Under conditions of certainty, accurate, measurable, and reliable information and knowledge on which you base your decisions are available to you. The cause-and-effect relationships are known. The future and outcome are highly predictable under conditions of certainty.

In essence, whereas uncertainty can stimulate processing and create a desire for information, certainty helps give an attitude durability and impact.

11. What are the limitations of Certainty Factor (CF)?

The formula for certainty factor (CF) depends on the number of premises in a rule:

Single premise: CF(H, E) = CF(E) * CF(rule) = CF(user) * CF(expert)

Multiple premises: CF(A AND B) = Minimum (CF(a), CF(b)) * CF(rule)

Multiple premises: CF(A OR B) = Maximum (CF(a), CF(b)) * CF(rule) 

 

CF is a clinical parameter that indicates how much confidence is given to MYCIN. The value of each premise or symptom is provided by an expert or literature. 

 

CFs have some advantages, such as mimicking how people handle numeric measures of uncertainty and providing efficient inference. However, they also have some disadvantages, such as treating all evidence as independent and overstating the joint contribution of related evidence. 

 

12. Conclusions

Certainty factors (CF) express belief in an event (or fact or hypothesis) based on evidence (or the expert's assessment), along a scale, say from 0 to 10, where 0 means false and 1 means true. These certainty factors are not probabilities. The certainty factor indicates how true a particular conclusion is.

13. FAQs

Q. What is reasoning using certainty factors in AI?

Ans. 

The certainty factor has a value between −1.0 and +1.0, where a negative 1.0 value indicates that the assertion can never be true in any scenario and a positive 1.0 value indicates that the statement can never be wrong.

Q. What's Emycin certainty factor combination functions in Lisp

Ans.

Here are the Emycin certainty factor combination functions in Lisp:

(defun cf-or (a b)

 "Combine the certainty factors for the formula (A or B).

 This is used when two rules support the same conclusion."

 (cond ((and (> a 0) O b 0))

 (+ a b (* -1 a b)))

 ((and (< a 0) (< b 0))

 (+ a b (* a b)))

 (t (/ (+ a b)

 (− 1 (min (abs a) (abs b)))))))

(defun cf-and (a b)

 "Combine the certainty factors for the formula (A and B)."

 (min a b))

References