Docsity
Log inSign up
Docsity
Prepare for your exams
Prepare for your exams

Study with the several resources on Docsity

Earn points to download
Earn points to download

Earn points by helping other students or get them with a premium plan

Guidelines and tips
Guidelines and tips
Sell on Docsity
Sell on Docsity
Log inSign up

Prepare for your exams

Study with the several resources on Docsity

Find documents

Prepare for your exams with the study notes shared by other students like you on Docsity

Search Store documents

The best documents sold by students who completed their studies

Search through all study resources
Docsity AINEW

Summarize your documents, ask them questions, convert them into quizzes and concept maps

Explore questions

Clear up your doubts by reading the answers to questions asked by your fellow students

Earn points to download

Earn points by helping other students or get them with a premium plan

Share documents
download point icon20 Points

For each uploaded document

Answer questions
download point icon5 Points

For each given answer (max 1 per day)

All the ways to get free points
Get points immediately

Choose a premium plan with all the points you need

Study Opportunities

Choose your next study program

Get in touch with the best universities in the world. Search through thousands of universities and official partners

Community

Ask the community

Ask the community for help and clear up your study doubts

University Rankings

Discover the best universities in your country according to Docsity users

Free resources

Our save-the-student-ebooks!

Download our free guides on studying techniques, anxiety management strategies, and thesis advice from Docsity tutors

From our blog

Exams and Study
Go to the blog

Understanding Rule-Based Expert Systems: Principles and Components, Exams of Reasoning

Davao Doctors College (DDC)Reasoning

An introduction to rule-based expert systems, a type of computer program that uses domain knowledge for problem-solving. Expert systems have been applied in various fields such as medicine, geology, chemistry, and computers. They act as advisors, perform tasks requiring human expertise, and use symbolic processing and explanation. Rules, the backbone of expert systems, are conditional statements that establish conditions for actions.

Typology: Exams

2021/2022

Uploaded on 08/01/2022

hal_s95
hal_s95 🇵🇭

4.4

(652)

10K documents

1 / 9

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
- 161 -
EXPERT SYSTEMS: AN OVERVIEW
Felisa Verdejo
Facultad de Informática, Universidad del País Vasco, Spain.
ABSTRACT
The purpose of this article is to introduce readers to the basic
principles of rule-based expert systems. Four topics are discussed
in subsequent sections: (1) Definition; (2) Structure of an expert
system; (3) State of the art and (4) Impact and future research.
l.WHAT
IS AN EXPERT SYSTEM?
The effort of many years of research in Artificial Intelligence (A.I.) has led to
practical results. One of the most promising fields has come to be known as expert
systems.
An expert system is a computer program that solves problems by using specific
domain knowledge-based reasoning. Expert systems have been developed mainly in areas such
as medicine, geology, chemistry, and computers. They act as advisors on problems
requiring judgmental knowledge to combine the relevant factors leading to a decision.
Characteristics that define an expert system are:
1. Expertise: an expert system performs tasks normally requiring significant human
expertise in an interactive mode with the user.
2.
Symbolic processing: in an expert system, knowledge is represented in symbolic form
and manipulated by inference mechanisms involving uncertainty.
3. Explanation: an expert system is able to explain its way of reasoning (how and why
it reachs
conclusions).
4.
Flexibility: an expert system acquires knowledge incrementally by increasing the
knowledge base it becames more expert.
Some of the well-known expert systems are:
DENDRAL-'-) - deduces a chemical structure from a formula and its corresponding
mass spectogram.
MYCIN2) - diagnoses patients with bacterial infections.
PROSPECTOR3) - aids in problems of mineral exploration to determine the most
favorable drilling sites.
XCON4) - configures computer systems from orders, insuring that they are
complete and specifying the spatial relationships among components.
SACON^) - selects an appropiate analysis strategy for studying a case with the
MARC system (a computer program to simulate a mechanical structure
behavior to various load
conditions).
DIPMETER ADVISOR6) - interprets oil field logs to decide where to drill wells.
2.STRUCTURE OF AN EXPERT-SYSTEM
It is widely recognised that the source of a human expert's performance is the domain
specific knowledge that he
uses.
Conventional programs encode knowledge as algorithms. A
different approach is to express the task knowledge in a declarative form separated from
the control component. Several schémas have been proposed, here we will mention two of them:
. Predicate calculus. The information is represented as assertions and manipulated by
theorem proving techniques. PROLOG, a general purpose computational language with an
inference procedure based on the resolution principle is a well known example.
pf3
pf4
pf5
pf8
pf9

Partial preview of the text

Download Understanding Rule-Based Expert Systems: Principles and Components and more Exams Reasoning in PDF only on Docsity!

EXPERT SYSTEMS: AN OVERVIEW

Felisa Verdejo

Facultad de Informática, Universidad del País Vasco, Spain.

ABSTRACT The purpose of this article is to introduce readers to the basic principles of rule-based expert systems. Four topics are discussed in subsequent sections: (1) Definition; (2) Structure of an expert system; (3) State of the art and (4) Impact and future research.

l.WHAT IS AN EXPERT SYSTEM? The effort of many years of research in Artificial Intelligence (A.I.) has led to practical results. One of the most promising fields has come to be known as expert systems. An expert system is a computer program that solves problems by using specific domain knowledge-based reasoning. Expert systems have been developed mainly in areas such as medicine, geology, chemistry, and computers. They act as advisors on problems requiring judgmental knowledge to combine the relevant factors leading to a decision. Characteristics that define an expert system are:

  1. Expertise: an expert system performs tasks normally requiring significant human expertise in an interactive mode with the user.
  2. Symbolic processing: in an expert system, knowledge is represented in symbolic form and manipulated by inference mechanisms involving uncertainty.
  3. Explanation: an expert system is able to explain its way of reasoning (how and why it reachs conclusions).
  4. Flexibility: an expert system acquires knowledge incrementally by increasing the knowledge base it becames more expert. Some of the well-known expert systems are: DENDRAL-'-) - deduces a chemical structure from a formula and its corresponding mass spectogram. MYCIN^2 ) - diagnoses patients with bacterial infections. PROSPECTOR^3 ) - aids in problems of mineral exploration to determine the most favorable drilling sites. XCON^4 ) - configures computer systems from orders, insuring that they are complete and specifying the spatial relationships among components. SACON^) - selects an appropiate analysis strategy for studying a case with the MARC system (a computer program to simulate a mechanical structure behavior to various load conditions). DIPMETER ADVISOR^6 ) - interprets oil field logs to decide where to drill wells. 2.STRUCTURE OF AN EXPERT-SYSTEM It is widely recognised that the source of a human expert's performance is the domain specific knowledge that he uses. Conventional programs encode knowledge as algorithms. A different approach is to express the task knowledge in a declarative form separated from the control component. Several schémas have been proposed, here we will mention two of them: . Predicate calculus. The information is represented as assertions and manipulated by theorem proving techniques. PROLOG, a general purpose computational language with an inference procedure based on the resolution principle is a well known example.

. Ruled-based systems. Domain knowledge is expressed in a uniform way by means of rules and separated from the procedures that will use it. Most of the expert systems developed so far are rule based. The emphasis in this framework is not on the theoretical aspects but on its conceptual adequacy, applicability, and performance. In this paper we will focus on the rule-based approach.

2.1 Rules Roughly speaking a rule is a conditional statement either relating facts or establishing the conditions to perform some actions. One example of rule relating facts is the SACON rule outlined below "IF the material composing the sub-structure is one of the metals, AND the tolerable analysis error is between 5 and 30, AND the non-dimensional stress is greater than 0,9, AND the number of loading cycles to be applied is between 10 and 100 THEN fatigue is one of the stress behavior in the sub-structure." A representative rule of the form "condition -> action," extracted from XC0N follows : "IF the current subtask is assigning devices to unibus modules AND there is an unassigned dual port disk drive AND the type of controller it requires is known AND there are two of such controllers, neither of which has devices assigned to it AND the number of devices which these controllers can support is known THEN assign the disk drive to each controller AND note each controller supports one device."

2.2 Knowledge base

A knowledge base is formed by the set of rules that synthesize the body of human expertise in the problem domain. Some rules would express causal interconnection among facts with a true value in the logical sense. Others would suggest what knowledge is relevant to a class of problems. Some other rules would represent likely relations observed among phenomena involving words such as "often" or "sometimes". These criteria learned in general by experience and practise, are called rules of thumb or heuristics in A.I. terminology. When used in reasoning about a problem, they are manipulated with a degree of confidence. Conclusions are reached by making conjectures and increasing their evidence. The final advice includes a measure of belief for the suggested interpretation. 2.3 Data base The data base of an expert system contains the facts (known or inferred) about the current case being studied. The following are examples of facts: The material of the substructure is one of the metals. The tree has needles bundler together. J. Smith has a shiff neck. Starter turns engine slowly or not at all. There is a type-A porphyry copper deposit (2.65)

    • larch

I

I

western larch-| ' - - needles three sided

"Tree is a larch" and "needles three sided" are again hypotheses (sub-goals) to be determined. Subgoals are tried succesively. In our exemple, to prove the hypothesis "tree is a larch", rule 3 might be selected.

    • conifer

I

I

larch - - - - - - neddles bundled toghether

I

^- - bundle more than six needles

At this stage rule 2 couH be applied. The system would ask the user "Has the tree needles or scales-like leaves?" After receiving confirmation the process continues until there are no more subgoals to attemp. When they are proven, the goal is then said to be concluded. Backward chaining strategy is goal oriented. It begins with a given hypothesis and looks for the conditions to be satisfied. The user is asked questions only if the system can no longer infer information from its knowledge base. B) Forward chaining. Rules have one of the two forms: IF THEN Example: IF starter turns engine slowly or not at all THEN perform headlight test IF THEN Example: IF there is voltage at the battery AND there is no voltage at the condenser THEN •the wire between the battery and the condenser is broken, and the interpreter executes as follows

match the left-hand side (Situation/Premise) whitin the data base if the match is succesful the rule might be selected to apply the rule means

. to perform the action part possibly modifying the content of the data base (first form) . to add the consequent as a new fact. Actions in the second form are restricted to adding facts to the data base. Forward chaining strategy is data-driven. It begins with the data provided by the user, selecting rules by pattern-matching and executing the corresponding actions. Let us emphasize that systems based on rules following the pattern: IF THEN are deduction systems and can be interpreted either forward, or backward. Choosing a mode depends on the structure of the domain and the objectives pursued. As we have seen, the interpreter matches the content of the data base with the rules. In the case that several rules are satisfied, a mechanism to decide which one should be applied is needed. Among the different criteria that have been used are the following:

. Execute the first satisfied rule. . Express the way of selecting by rules (meta-rules). . Consider first the most/least used rule. . Select the rule with the larger number of facts (the more specific). . All the rules will be tried. By now we have introduced the basic architecture of a rule based expert system, here is a summary:

  1. A knowledge base of rules capturing the expertise of a domain.
  2. A database of facts containing the information of the current case.
  3. An inference procedure with: a. a matching mechanism to compare the content of the database with the rules, b. a mode of inference, backward or forward chaining, c. a conflict resolution strategy to select among the satisfied rules. The system performs by executing cycles formed by three steps : . Recognize (by matching) the candidate rules to apply, . Select a rule (following the conflict strategy), . Apply the rule and continues until a solution is reached or there are no more rules to apply. The separation of knowledge from control and the expression of knowledge by means of rules are the main features that make this framework useful. The first one enables us to use the same inference procedure for different domains; the second offers -but does not guarantee- an extensibility so that rules can be added/deleted without affecting the other components. This allows designers to develop large systems incrementally. 2.5 Explanation There is an additional benefit to structuring knowledge in form of rules : the reasoning process becomes transparent and provides the system with an explanatory power. The user can ask questions about WHY a fact was or not considered, and HOW a conclusion was reached. The answers are based on the rules currently being applied. Following is an example from MYCIN: System - Was penicillinase added to this blood culture? User - WHY (do you ask this question?) System - This will aid in determining wether organism-1 is a contaminant. It has already been established that . The site of the culture-1 is blood . The gram stain of organism-1 is grampos Therefore if penicillinase was added to this blood culture, then is weakly suggestible evidence (.4) that organism-1 is a contaminent. The required answer is extracted from the rule the system is working on
    • penicillinase added

I

I

Organism-1 is a contaminent (.4) - - - - site is blood

I

l_ _ gram-stain is grampos

. Potential high benefits (intellectual or economical) due to the cost of building the knowledge base ; . Lack of appropiate hardware at reasonable cost. Prices are dropping but machines allowing highly efficient symbol processing are still expensive. Over the past decade a broad range of expert systems have been built to solve problems in different areas. The following list illustrates some of the applications: . Fault diagnosis . Medical diagnosis . Chemical data interpretation . Oil well log interpretation . Mineral exploration . Electronic troubleshooting . Signal interpretation . Speech undertanding . Computer configuration . Computer aid instruction . Planning experiments . Electronic design. Different attempts have been made to identify among the applications which are the generic tasks that would specify what an expert system does. A first characterization is to distinguih between analysis and synthesis. The analysis task consists of the identification or interpretation of data. Most of the systems are related to this task. What they do is classification problem solving, providing an answer from a preenumarated set of solutions. Problems of synthesis, such as planning or design, require that a solution be built up, usually verifying a number of constraints. These problems are difficult, but important research is going on and some significant results are being obtained. Often a domain task might be decomposed into a number of parts (either analytic or synthetic). An expert system for a practical application can combine both approaches. 3.2 Tools

An adequate software tool is a factor of great practical importance. Not only can it reduce the time of development but it can also help the designer to focus on the task- specific knowledge allowing him rapid prototyping to explore appropiate solutions. Several possibilities are offered:

  1. A selection of a rule-based framework such as EMYCIN or OPS, tools evolved from previous expert-systems and abstracted to being used in other domains. They usually provide . A language to express rules . Editing and debugging tools to create and maintain the knowledge base . An inference mechanism . An explanation module . A consultation interface. This option is worth in the cases where the structure of the new problem (generic tasks) fits the specification of one of the frameworks. For a detailed comparaison of these tools see"').
  1. There is a class of tools, (KEE, AGE, LOOPS..) more general than the previous one, that has no comittements to a particular form of rules or mode of inference. It allows to define different models. The price for this flexibility is more complexity. These languages (built on top of LISP) are suitable for advanced users or research work.
  2. To use an artificial intelligence programming environment such as INTERLISP op POPLOG and write an expert-system beginning from the scratch. This way takes much more time. In any case, as Davis points out, developping a substantial expert-system with real performance takes at least five man-years of effort, assuming the team has some background in A.I. 3.3 Implementation methodology The process of building a serious expert-system is an empirical activity which requires two kinds of participants, human experts, who know about the problem, and so-called knowledge enginners who have to design the appropiate architecture as well as produce an initial knowledge base for the system. Transfering expertise from human to computers is recognized as a hard work, with an important investment of time and manpower. The main effort lies in the creation and maintenance of the knowledge base for the specific domain. Systems already developed suggest the following constructions stages : Stage 1-producing a prototype. Main steps are: . Problem definition: to identify the task, fix the role of the system and establish a common language between the domain expert and the knowledge engineer. . Initial system design: to choose the knowledge representation and problem solving techniques. . Knowledge acquisition: to extract knowledge from human experts and express it in .a suitable form to be transferred to the system. . Evaluation: to measure the behavior of the system in common situations. Stage 2-Transfering the system to the operational environment. . Extensive development of the knowledge base with special attention to cover ordinary and uncommon situations. . Training the people to support the continuing improvement of the system. Stage 3-Getting the system to work. Development,however, will never end due to changes in the problem domain. The whole process can be characterized as one of progressive deepening. As the designers go on building a system, their undestanding of the task grows, so do the knowledge base and performance of the system.

4.IMPACT AND FUTURE RESEARCH Problems that resisted conventional software solutions have been sucessfully solved by expert-systems technology. This new approach has increased industrial expectation because of the high profits of replicating human expertise, specially in areas where there is a shortage of skilled personnel. For example, XCON (currently used by DEC corporation) has been saving the company 10 millions dollars a year. Artificial Intelligence has been characterized as the study of ill-defined problems to be transformed into defined ones. The attempt to formalize a domain, clarifying progressively the principles on which decisions are made, has also significant side-