Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 67

Lecture 7

Expert Systems
This week
 Artificial Intelligence (AI)
 Expert Systems (ES)
 Expertise, experts and transferring expertise
 Some knowledge elicitation techniques
 Knowledge representation
 Use of ES, examples
 Structure of ES
 Benefits
 Limitations
 Contrast with DSS
Artificial Intelligence (AI)

 Simulation of human intelligence by machines


 Characteristics
◦ learning
 acquisition of information and rules for using the information
◦ reasoning
 using the rules to reach approximate or definite conclusions
◦ self-correction
 updating existing material in response to feedback
 Applications include expert systems, tutoring
systems, speech recognition, machine vision
Expert Systems (ES) - summary

 Expert System
a decision-making computer package intended to
attain or surpass the levels of performance of a
human expert in some highly specialised field
Expert Systems (ES) - summary
 Expert System
 expertise of a human expert is transferred to ES
 knowledge is stored and can be accessed as required
 makes inferences and arrives at a specific conclusion
 differs from DSS in that it identifies the best solution to a
problem rather providing a range of possible solutions
 advises the user
 consists of
 knowledge base
 inference engine
 user interface
Expertise

 “Expertise
is the extensive, task-specific
knowledge acquired from
◦ training
◦ reading
◦ experience”
Expertise
 Includes
 Theories
 Rules and procedures
 Rules (heuristics) about what to do in a given situation
 Global strategies for problem-solving
 Meta-knowledge (knowledge about knowledge)
 Facts
Turban (2001)
 Recognise two main types
 Facts
 Rules (procedures)
Categorisation: novice to expert

 Barfield (1986) distinguished between


◦ experts
◦ non-experts
 naive
 novice
 intermediate
Categorisation: novice to expert

 Dreyfus (1987) suggested stages


◦ Novice
◦ Advanced beginner
◦ Competence
◦ Proficiency
◦ Expertise
Categorisation: novice to expert

 Dreyfus’s stages
1. Novice
o skill acquisition
o novice receives instruction
o rules followed blindly
o lacks coherent sense of task
o judges performance on use of rules
Categorisation: novice to expert

 Dreyfus’s
stages
2. Advanced beginner
◦ performance becomes barely acceptable after
experience of real situations
◦ recognises situations on perceived similarity to
previous examples
Categorisation: novice to expert

 Dreyfus’s
stages
3. Competence
◦ increased experience
◦ recognises similarities
◦ views decision-making in a hierarchical manner
◦ improves performance
◦ choice of plan is not a simple matter
Categorisation: novice to expert

 Dreyfus’sstages
4. Proficiency
◦ view altered by outcome of recent events
◦ certain features stand out - others ignored
◦ plan modified as necessary
◦ understanding the task becomes intuitive
 triggered naturally without explicit thought
Categorisation: novice to expert

 Dreyfus’sstages
5. Expertise
◦ proceeds without detached deliberation
◦ acts without conscious contemplation of options
◦ performance is fluid
◦ not interrupted by detached deliberation or analysis
Categorisation: novice to expert

 Kolodner (1983) distinguishes between


experts & novices

◦ experts have more knowledge of domain

◦ experts apply and use knowledge


more effectively than novices
Categories of expertise

 Clare (1989) categorised types of expert


◦ Practitioner

◦ Academic
◦ Craftsman

◦ Samurai
Categories of expertise

 Practitioner
◦ expertise based on organised principles and methods
◦ expertise gained from
 theoretical knowledge
 developed from experience and practice
◦ work environment may not offer fully coherent basis for
development of expertise
 such expertise is therefore rare
Categories of expertise

Academic
◦ individual expected to
 guide
 direct
 teach
others in field of own expertise
Categories of expertise

 Craftsman
◦ expertise gained in routine activities
◦ individual solves similar problems
◦ rote learning of methods and procedures
Categories of expertise

Samurai
◦ characterised by performance of expertise
◦ performance is the key aspect
◦ final outcome is consequence of performance
"There is definitely a moment ... when there is loss of self. ... you
are 'in the zone', when you are not aware of what you're
doing...you are simply in the act of doing it." Michael Boyd

"...it's similar with music in a way – although you have to learn the
notes first, you then have to forget the notes to play the music."
Oliver Sacks
extracts from "More than merely the sum of their parts", public discussion at Columbia
University, reported in The Sunday Times Culture Supplement, page 10-11, 23 November
2008
Knowledge acquisition

 Knowledge acquisition techniques extend and


enhance those used in systems analysis, such as
◦ repertory grid analysis
 structured interviews
 triads
to extract knowledge of relationships and
differences between elements
Stages of knowledge acquisition

Reformulations

Redesigns

Refinements

requirements concepts structure rules

Identify Find Design Formulate Validate


concepts to structure to rules to rules that
problem
represent organise embody organise
characteristics knowledge knowledge knowledge knowledge
Identification Conceptualisation Formalisation Implementation Testing

Buchanan, Barstow, Bechtal, Bennett, Clancey, Kulikowski, Mitchell and Waterman, Constructing an Expert
System: in Building Expert Systems (1983), ed. Hayes-Roth F, Waterman DA and Lenat DB.
Types of Knowledge to be Represented in a Knowledge Base

Behaviour Meta-knowledge Uncertain


descriptions (knowledge about facts
and beliefs knowledge)
Processes
Vocabulary
definitions
Constraints
Objects &
relationships Knowledge Facts about
Base the domain

Heuristics &
decision rules Disjunctive
facts
problem solving
procedures
General
Hypotheses knowledge
Typical
(theories)
situations

Adapted from: Fikes and Kehler 1985, in Turban


 ‘Non-experts’ often outnumber ‘experts’ by 100:1
 Top 10% of experts perform better:
◦ 3 times as well as ‘average’ experts
◦ 30 times better than those in bottom 10%
 Suggests that
◦ overall effectiveness can be improved by making top-level
expertise available to less expert decision-makers
Turban (2001)
 This is the aim of Expert Systems
 Expert activities
◦ Recognising and formulating the problem
◦ Solving problem quickly and correctly
◦ Explaining the solution
◦ Learning from experience
◦ Restructuring knowledge
◦ Breaking rules if necessary
◦ Determining relevance
◦ Being aware of limitations
Turban (2001)
Expert Knowledge Expert User
engineer system

Development Consultation
(including knowledge acquisition)
 Objective to transfer expertise from expert to ES to non-expert (user).
Involves:
◦ Knowledge acquisition
◦ Knowledge representation
◦ Knowledge inferencing
◦ Knowledge transfer to user Turban (2001)
 Repertory Grid Analysis
◦ Based on Kelly’s Personal Construct Theory
 every person has a personal view of the world, forms
theories, tests hypotheses, analyses results
◦ Structured interviews
◦ Triads
Structured interviews

 Expert is interviewed
◦ identifies important items in the domain
◦ identifies associated attributes
◦ creates a scale of characteristics (opposites)
◦ places items on scale
◦ solutions are placed on grid (table)
◦ ratings from scale may be reviewed
Repertory Grids (RGA)
 Generated following structured interview
◦ shows relative position of elements
◦ based on scoring in structured interview

Attributes Availability Calories Vitamins


Trait High (5) High (5) High (5)
Opposite Low (1) Low (1) Low (1)

Choc mints 5 5 1
Muesli bar 5 4 3
Carrot 4 2 4
Triads

 Groups of three items are classified


◦ why two are alike, and the third different
◦ combine in many different ways

◦ Consider:
 Margaret Thatcher
 Ronald Reagan
 Ed Miliband
Triads example
left-wing deceased female
Ed Margaret
Ronald
Miliband Thatcher
Reagan

Margaret Ed Ronald Ed
Margaret Ronald
Thatcher Miliband Reagan Miliband
Thatcher Reagan

right-wing living male


 When knowledge has been identified, it must be
stored and represented for use
◦ Semantic networks
◦ Association lists (alists)
◦ Search trees
◦ Frames
◦ Facts
◦ Rules
Semantic networks

 Items linked
◦ “is-a” relationships
◦ uses inheritance

has Is-a employee Is-a


office manager staff

Is Works-in
45 years old Jim sales
 Association list has attribute.value pairs:
((1 . “one”)(2 . “two”)(3 . “three”))

 this alist shows the names of numbers


 extracting the name of 1 will give “one”
 first attribute is used as key to extract value
 clumsy for large amounts of data
 Data can be stored
Jim
on the tree
 Tree can be

searched to find data


 Here, tree shows
Pat Mary
organisation structure

Liz Alan Ben Dora


Search trees

 Depth first search


Jim
◦ from start point
◦ continues until
downward path
exhausted
◦ then next downward Pat Mary
path is tried…
◦ until all paths taken

Liz Alan Ben Dora


Search trees

 Breadth first search


Jim
◦ from start point
◦ considers all options at
next level down
◦ then moves to
subsequent level… Pat Mary
◦ until all levels visited

Liz Alan Ben Dora


 If tree becomes too
Jim
large it is impossible
to search the entire
tree (time constraints)
 Unproductive paths
Pat Mary
must be pruned
 If searching for

Mary’s staff, ignore


staff working for Pat Liz Alan Ben Dora
 Similar to a blank form (standard template)
 stereotypes used, other values specified

Frame: SALARY
rate of pay: unit (£ per annum)
tax paid to date: unit (month, year)
calculate ((TAX-PAID) (tax))
NI deductions: unit (calculate (NI-CON))
pension contributions: unit (calculate (PENSION))
 Data that is known to be true

 prolog can store data as facts in database

manages(jim, pat).
manages(jim, mary).
manages(pat, liz).
manages(pat, alan).
manages(mary, ben).
manages(mary, dora).
 Production rules
◦ format:
 pattern action
 typically in the form of if…then…
 rules in prolog:
predicate is true if components satisfied
sibling(X,Y) :- parents(X, M, F), parents(Y, M, F).
 MYCIN
◦ medical diagnosis, identifying treatment for blood
disorders
◦ symptoms entered
◦ ES compares with known symptoms
 presents diagnosis
 prompts for further information
◦ each rule has a certainty factor
◦ backward chaining mechanism

Chaffey (2003)
Use of ES: examples (2)

 Prospector
 Evaluates mineral potential of geological area
 Assessing risk of investments
 uses probabilistic reasoning (based on Bayes’ theorem)
 can explain why it needs answers to specific questions
 will choose new hypotheses as necessary
Use of ES: examples (3)

 XCON
◦ configures VAX computer systems for DEC
◦ rule-based expert system
◦ probabilistic information not necessary
◦ exact statement of requirements in each case
Use of ES: examples (4)

 CLUES
Countrywide’s Loan Underwriting Expert System
 developed to ensure rapid, consistent, high-quality loan decisions
 CLUES has 400 rules
 refined until it agreed with 95% of ‘human expert’ decisions
 human underwriter still examines all rejected loans
 productivity increased from 6 or 7 applications/day to 16/day

(Laudon & Laudon, 2004)


Use of ES: examples (5)

 CLUES Credit / loan decisions


 Customer provides personal details
◦ employment history, residence, etc.
 ES makes assessment of credit risk
◦ could be based on rules
 e.g. if loan is > 10% of salary… then… etc.
◦ or more advanced
 e.g. based on pattern of risk of similar customer profiles
 requires use of other techniques e.g. neural networks

(Laudon & Laudon, 2004)


Structure of ES (1)

 Main components
◦ Knowledge base
◦ Inference engine
◦ User interface
Structure of ES (2)

 Other components
 Knowledge acquisition sub-system
 Blackboard (workspace)
 Explanation facility (justifier)
Structure of ES (3)

 ‘Human’ components
 Domain expert
 Knowledge engineer
 User
Knowledge base

 The software that represents the knowledge


◦ Facts: what is known about the problem area (theories,
details of the problem situation, etc.)
◦ Rules: rules of inference, logical deductions, if… then…
etc.

Knowledge base
Knowledge Facts User
acquisition and
Rules Inference
engine
Inference engine

 The ‘brain’ of ES, ‘control structure’, ‘rule interpreter’


 Enables reasoning based on
 knowledge base (facts and rules in problem domain)
 blackboard (input about the particular problem to be solved)
◦ forward chaining (data driven)
 follows the rules to arrive at a conclusion
◦ backward chaining (goal driven)
 starts with hypothesis to be confirmed or disproved

Inference engine
Interpreter
Scheduler
Consistency enforcer
Knowledge Blackboard
base (workspace)
User interface

 ES user interface contains a language processor

 Consultation best carried out in natural language,


supplemented by menus, forms and graphics
Other components (1)
 Knowledge acquisition sub-system
 accumulation
 transfer
 transformation
of expertise from experts or documented sources
to the knowledge base

 process often supported by knowledge engineer


Other components (2)
 Blackboard (workspace)

 area of working memory for the description of the


current problem as specified by input data

 also used for recording


 plans (how to solve)
 agendas (actions awaiting)
 solutions (candidate hypotheses/solutions so far)
Other components (3)
 Explanation facility (justifier)

 in simple ES
 shows the rules used to derive solution

 in complex cases answers the questions


 How was this conclusion reached?
 Why was that one rejected? ...
‘Human’ components (1)

 Domain expert

◦ qualities already discussed

◦ may be more than one expert


 problems arise if they disagree!
‘Human’ components (2)
 Knowledge engineer

 helps the expert structure knowledge of domain

 effectively builds the ES / knowledge base

 in short supply
 needs some knowledge of domain
 needs greater knowledge of development environment
‘Human’ components (3)
 Different categories of users
 non-expert
 uses ES as a consultant or advisor
 student
 uses ES as an instructor
 ES builder
 uses ES to improve the knowledge base
 uses ES as a partner
 expert
 uses the ES as a colleague or assistant
e.g. for a second opinion
Benefits of ES (1)

 Monetary savings
◦ fewer human experts needed

 Improved quality of decisions / solutions


◦ ES are more consistent, fewer mistakes are made
◦ ES do not “forget” the rules

 Compatible with various decision styles

 Use as a training vehicle


◦ less time than gaining expertise “the hard way”
Benefits of ES (2)

 Expert is freed from repetitive, time consuming tasks

 Scarce expertise is preserved

 Ability to operate in hazardous environments


Limitations of ES (1)

 Knowledge is not always readily available


 Expertise can be hard to extract
from humans
 The approach of different experts
to situation assessment may be
different, yet correct
 It is hard for experts to abstract good situational
assessments when under time pressure
Limitations of ES (2)

 Users of ES have natural cognitive limits

 ES work well only in a domain of knowledge

 Most experts have no independent means of


checking whether their conclusions are reasonable
Limitations of ES (3)

 The vocabulary (jargon) experts use to express


facts and relationships is often limited and not
understood by others outside that field of
interest

 Help is often required from knowledge engineers


who are rare and expensive - can make ES
construction costly
Turban (2001)
Limitations of ES (4)

 Lack of trust by end-users


 may be a barrier to ES usage

 Knowledge transfer is subject to many perceptual


and judgmental biases
Turban (2001)
Contrast ES with DSS

Expert Systems DSS


 highly specialised in nature  extend range, capability and
effectiveness of managers' decision-
making processes
 extend scarce resource
(expertise)  applied where analytic aids of value
but manager's judgement essential
 allow non-experts access to
knowledge of experts  do not attempt to automate the
decision process, or impose
solutions
 aim to provide specific solution

 require specialist development


skills and tools
References and suggested reading

 Barfield, W, 1986, Expert-Novice Differences for Software:


Implications for Problem-Solving and Knowledge Acquisition,
Behaviour and Information Technology, Vol. 5, No. 1, p. 15-29
 Chaffey, D. (ed.), 2003, Business Information Systems, 2nd ed., FT
Prentice Hall
Expert Systems: chapter 6, pages 255-257
 Clare, J, 1989, Knowledge Elicitation for Financial Dealers p.237-
246 in Diaper D (editor) Knowledge Elicitation: Principles,
Techniques and Applications, Ellis Horwood
 Dreyfus, HL, 1987, Misrepresenting Human Intelligence p. 41-54 in
Born RP (editor) Artificial Intelligence: The Case Against, Croom
Helm
References and suggested reading
 Kolodner, JL, 1983, Towards an Understanding of the Role of
Experience in the Evolution from Novice to Expert, IJMMS 19,
p497-518
 Laudon, K. & Laudon, J., 2004, Management Information Systems,
8th ed., Pearson Prentice Hall
Expert Systems: chapter 10, pages 327-333
 Turban E. & Aronson J.E., 2001, Decision Support Systems and
Intelligent Systems (6th edition), Prentice Hall Business Publishing
Expert Systems: chapter 10, pages 407 - 428
 Whiteley, D., 2004, Introduction to Information Systems, Palgrave
Expert Systems: chapter 11, pages 161-168
 http://www.enquirewithin.co.nz/ (accessed 14 November 2006)

You might also like