EXPERT SYSTEMS: HISTORY, CHARACTERISTICS, ADVANTAGES, DISADVANTAGES - TECHNOLOGY - 2024

The expert systems are defined as systems that emulate the decision making ability of a human expert in a particular field. They use both heuristic strategies and facts to solve complex decision-making problems reliably and interactively.

They are designed to solve highly complex problems, reasoning through knowledge bases. Rather than being represented by procedural code, they are basically represented by If-Then rules.

Source: pixabay.com

They are able to express themselves and reason about some area of ​​knowledge, which allows them to solve many problems that would generally require a human expert. Expert systems were the predecessors of today's artificial intelligence, deep learning, and machine learning systems.

An expert system cannot substitute for a worker's overall performance in troubleshooting. However, they can drastically reduce the amount of work the individual must do to solve a problem, leaving the creative and innovative aspects of problem solving to people.

They have played an important role in many industries, such as financial services, telecommunications, healthcare, customer service, video games, and manufacturing.

System capacity

An expert system incorporates two subsystems: a knowledge base, which contains accumulated facts and experience, and an inference engine, which is a set of rules to apply to the knowledge base or known facts in each particular situation, in order to deduce new ones. acts.

System capabilities can be enhanced with additions to the knowledge base or rule set.

For example, today's expert systems may also have the ability to learn automatically, allowing them to improve their performance based on experience, just as humans do.

In addition, modern systems can more easily incorporate new knowledge and thus be easily updated. Such systems can better generalize from existing knowledge and handle large amounts of complex data.

History

- Initial developments

In the late 1950s, experimentation began with the possibility of using computer technology to emulate human decision-making. For example, computer-aided systems began to be created for diagnostic applications in medicine.

These initial diagnostic systems entered patient symptoms and laboratory test results into the system to generate a diagnosis as a result. These were the first forms of expert systems.

- Main developments

At the beginning of the sixties, programs were developed that solved well-defined problems. For example, games or machine translations.

These programs required intelligent reasoning techniques to handle the logical and mathematical problems that were presented, but they did not require much additional knowledge.

Researchers began to realize that to solve many interesting problems, programs not only had to be able to interpret the problems, they also needed basic knowledge to fully understand them.

This gradually led to the development of expert systems, which were more focused on knowledge.

The concept of expert systems was formally developed in 1965 by Edward Feigenbaum, a professor at Stanford University, USA.

Feigenbaum explained that the world was moving from data processing to knowledge processing, thanks to new processor technology and computer architectures.

Dendral

In the late 1960s, one of the first expert systems, called Dendral, was developed addressing the analysis of chemical compounds.

Dendral's knowledge consisted of hundreds of rules that described the interactions of chemical compounds. These rules were the result of years of collaboration between chemists and computer scientists.

- Maturity

Expert systems began to proliferate during the 1980s. Many of the Fortune 500 companies applied this technology in their daily business activities.

In the 1990s, many business application vendors, such as Oracle and SAP, integrated the capabilities of expert systems into their suite of products, as a way of explaining business logic.

characteristics

- Level of Experience

An expert system must offer the highest level of expertise. Provides efficiency, precision and imaginative problem solving.

- Reaction on time

The user interacts with the expert system for a fairly reasonable period of time. The time of this interaction must be less than the time that an expert takes to obtain the most precise solution for the same problem.

- Reliability

The expert system must have good reliability. To do this, you must not make any mistakes.

- Effective mechanism

The expert system must have an efficient mechanism to manage the compendium of knowledge existing in it.

- Handle problems

An expert system must be able to handle challenging problems and make the right decisions to provide solutions.

- Components

Knowledge base

It is an organized collection of data corresponding to the scope of experience of the system.

Through interviews and observations with human experts, the facts that make up the knowledge base must be taken.

Inference engine

Interpret and evaluate the facts in the knowledge base through rules, in order to provide a recommendation or conclusion.

This knowledge is represented in the form of If-Then production rules: "If a condition is true, then the following deduction can be made."

Conclusions

Often a probability factor is attached to the conclusion of each production rule and to the final recommendation, because the conclusion reached is not an absolute certainty.

For example, an expert system for the diagnosis of eye diseases could indicate, based on the information provided, that a person has glaucoma with a probability of 90%.

Also, the sequence of rules through which the conclusion was reached can be shown. Monitoring this chain helps assess the credibility of the recommendation and is useful as a learning tool.

Types

Rule-based

In this system knowledge is represented as a set of rules. The rule is a direct and flexible way of expressing knowledge.

The rule consists of two parts: the “If” part, called the condition, and the “Then” part, called the deduction. The basic syntax of a rule is: If (condition) Then (deduction).

Based on fuzzy logic

When you want to express knowledge using vague words like "very little", "moderately difficult", "not so old", fuzzy logic can be used.

This logic is used to describe an imprecise definition. It is based on the idea that all things are described on a sliding scale.

Classic logic operates with two certainty values: True (1) and False (0). In fuzzy logic, all certainty values ​​are expressed as real numbers in the range 0 to 1.

Fuzzy logic represents knowledge based on a degree of truthfulness, rather than the absolute truthfulness of classical logic.

Neuronal

The benefits of the rule-based expert system also combine the benefits of the neural network, such as learning, generalization, robustness, and parallel information processing.

This system has a neural knowledge base, rather than the traditional knowledge base. Knowledge is stored as weights in neurons.

This combination allows the neural expert system to justify its conclusions.

Neuronal-diffuse

Fuzzy logic and neural networks are complementary tools for building expert systems.

Fuzzy systems lack the ability to learn and cannot adapt to a new environment. On the other hand, although neural networks can learn, their process is very complicated for the user.

Neural-fuzzy systems can combine the computing and learning capabilities of the neural network with the representation of human knowledge and the explanation skills of fuzzy systems.

As a result, neural networks become more transparent, while the fuzzy system becomes capable of learning.

Advantage

Availability

Expert systems are readily available, anywhere, anytime, due to mass production of the software.

Reduced risk

A company can operate an expert system in environments that are dangerous to humans. They can be used in any hazardous environment where humans cannot work.

Business knowledge

They can become a vehicle to develop organizational knowledge, in contrast to the knowledge of individuals in a company.

Answer explanation

They are able to give an adequate explanation of their decision making, expressing in detail the reasoning that led to an answer.

When used as training tools they result in a faster learning curve for beginners.

Fast answer

Helps to get fast and accurate answers. An expert system can complete its share of tasks much faster than a human expert.

Low error rate

The error rate of successful expert systems is quite low, sometimes much lower than the human error rate for the same task.

Emotionless response

Expert systems work without getting excited. They don't get tense, fatigued or panic, and they work steadily during emergency situations.

Knowledge permanence

The expert system maintains a significant level of information. This contained knowledge will last indefinitely.

Rapid prototyping

With an expert system, it is possible to enter some rules and develop a prototype in days, rather than the months or years commonly associated with complex IT projects.

Multiple experiences

The expert system can be designed to contain the knowledge of many qualified experts and thus have the ability to solve complex problems.

This reduces the expense of consulting expert problem solving consultants. They are a vehicle for obtaining sources of knowledge that are difficult to obtain.

Disadvantages

Knowledge acquisition

It is always difficult to obtain the time of experts in particular fields for any software application, but for expert systems it is especially difficult because experts are highly valued and constantly requested by organizations.

As a consequence, a large amount of research in recent years has focused on tools for the acquisition of knowledge, which help to automate the process of design, debugging and maintenance of the rules defined by experts.

System integration

The integration of the systems with the databases was difficult for the first expert systems, because the tools were mainly in languages ​​and platforms not known in corporate environments.

As a result, a great effort was made to integrate expert systems tools with legacy environments, making the transfer to more standard platforms.

These problems were mainly solved by the paradigm shift, as PCs were gradually accepted in the computing environment as a legitimate platform for the development of serious business systems.

Processing complexity

Increasing the size of the knowledge base increases the complexity of the processing.

For example, if an expert system has 100 million rules it is obvious that it would be too complex, and it would face many computational problems.

An inference engine would have to be able to process a large number of rules to make a decision.

When there are too many rules, it is also difficult to verify that these decision rules are consistent with each other.

It is also difficult to prioritize the use of the rules to operate more efficiently, or how to resolve ambiguities.

Knowledge update

One problem related to the knowledge base is how to make updates quickly and effectively. Also, how to add new knowledge, that is, where to add it among so many rules.

Applications

Diagnosis and troubleshooting

It summarizes all fault-inferring systems and suggests corrective actions for a malfunctioning process or device.

One of the first knowledge areas where expert systems technology was applied was medical diagnosis. However, engineering systems diagnostics quickly outperformed medical diagnostics.

The diagnosis can be expressed as: given the evidence presented, what is the underlying problem, reason or cause?

Planning and scheduling

These expert systems analyze a set of objectives to determine a set of actions that achieve those objectives, providing a detailed ordering of those actions over time, considering materials, personnel and other restrictions.

Examples include airline staffing and flight scheduling, and manufacturing process planning.

Financial decisions

Financial advisory systems have been created to help bankers determine whether to make loans to individuals and companies.

Insurance companies use these expert systems to assess the risk that the client presents and thus determine the price of insurance.

Process monitoring and control

They analyze data from physical devices in real time, in order to notice anomalies, predict trends and control both optimization and fault correction.

Examples of these systems are in the oil refining and steelmaking industries.

Knowledge consulting

The primary function of this application is to provide meaningful insight into the user's problem, within the environment of that problem.

The two expert systems that are most widely distributed throughout the world belong to this category.

The first of these systems is an advisor that advises the user on the correct use of grammar in a text.

The second is a tax advisor who is attached to a system for preparing taxes. Advises the user on the strategy and particular tax policies.

References

1. Guru99 (2019). Expert System in Artificial Intelligence: What is, Applications, Example. Taken from: guru99.com.
2. Wikipedia, the free encyclopedia (2019). Expert system. Taken from: en.wikipedia.org.
3. Margaret Rouse (2019). Expert system. Techtarget. Taken from: searchenterpriseai.techtarget.com.
4. Vladimir Zwass (2019). Expert system. Encyclopaedia Taken from: britannica.com.
5. Wtec (2019). The Applications of Expert Systems. Taken from: wtec.org.
6. Viral Nagori (2014). Types of Expert System: Comparative Study. Semantic Scholar Taken from: pdfs.semanticscholar.org.
7. World of Computing (2010). Expert Systems. Taken from: intelligence.worldofcomputing.net.