You are on page 1of 9

CHAPTER 7

KNOWLEDGE CODIFICATION

TEST YOUR UNDERSTANDING

1. What assumptions can be made about knowledge codification? Explain.

There are three assumptions about knowledge codification:

a. Accuracy in the facts, beliefs, perceptions, and assumptions affect knowledge


codification
b. Knowledge codification is meta-knowledge. This crucial assumption assumes that
there exists knowledge about the structure of stored knowledge, about objects,
relationships, and how to perform certain activities.
c. Modularity of the design. With interrelated modules, errors can be more easily
detected and corrected.

2. Distinguish between:
a. rule premise and action
b. slot and facet
c. decision table and decision tree

a. The premise is the yes/no, true/false statement that is to be evaluated; the action is the
event that will occur if the premise is true. Therefore, the premises lead to the action,
making up a production rule or a rule.
b. A slot is the specific object being described in a frame. A facet is the value of the
slot. For example, if the slot is airline, the facet may be United.
c. A decision table is a list of conditions with their respective values matched against a
list of conclusions. A decision tree is a hierarchically arranged semantic network that
is closely related to a decision table. It is composed of nodes that represent goals and
links that represent decisions or outcomes.

3. Two key elements make up a frame. Use an example of your own to illustrate these
elements.

A frame is a structure for organizing knowledge through previous experience that uses
slots and facets. Slots are the specific objects being described and facets are the values of
the slots. If the slot were airlines, the facets could be United, USAIR, American, Delta,
etc. Students might have examples unique to their areas of interest.

4. How can frames be linked? Explain.

Frames may be linked if one frame inherits the properties of a higher-level frame. If the
facet of one frame is the slot of another frame, a link occurs. For example, if in frame 1,
the slot is airline and the facet is United and in frame 2, the slot is United and the facet is
CHAPTER 7
KNOWLEDGE CODIFICATION

Boeing 777, the two frames are linked through United. Frame 2 would be the child slot,
while frame 1 is the parent slot.

5. Summarize the pros and cons of decision tables versus decision trees. Under what
conditions would you use one tool over the other?

The main advantage of a decision table is ease of construction and conversion to a set of
rules. They can also be used during rapid prototyping and testing to check the logic of
the knowledge-based system. The main drawback is that the mental steps in manually
creating, editing, and checking one makes decision tables unwieldy and difficult to
construct.

6. In your own words, write a short essay to explaining the importance of knowledge
codification.

Knowledge codification is important, because without codification, there can be no


knowledge sharing via electronics or technology. Once codified, a KM system may be
for diagnostic purposes, instruction and training, interpretation, planning and scheduling,
prediction, and other areas of value-added gestures to the continuing changes in a
learning organization. For students, there should be numerous sources of reference on
the Internet addressing knowledge codification.

7. In your own words, explain knowledge codification. How does it differ from
knowledge creation?

Knowledge codification is essentially organizing and coordinating knowledge in a form


and a structure meaningful to the user. Knowledge creation is generating experience via
teamwork, repositories, or other interfaces

8. Present justification for knowledge codification. How important a step does it play in
building knowledge management systems?

Organizations constantly adapt to changing situations—changes in market, changes in


customer tastes and preferences, turnover among company specialists and experts, and
new developments in technology. In codifying knowledge, the resulting knowledge base
serves in several important training and decision-making areas. See answer 6 above.

9. How important is planning in knowledge codification? Be specific.

A lack of planning is reflected in poor quality solutions, incomplete answers, or


inefficient systems. Planning ensures proper transition from knowledge capture to
knowledge codification, verification, validation, and implementation. A knowledge
developer is responsible for planning the development and execution phases and how the
user will interface with the resulting KM system—activities that are difficult to carry out
without a plan.

Dr. Faouzi Maddouri 7-2


CHAPTER 7
KNOWLEDGE CODIFICATION

10. In what way(s) do the four modes of knowledge conversion result in socialization and
externalization?

• From tacit to tacit knowledge produces socialization, where experience is what a


knowledge developer looks for in knowledge capture. Observation and practice are
two important knowledge capture tools.
• From tacit to explicit knowledge is externalizing, explaining, or clarifying tacit
knowledge via analogies, models, or metaphors. Explicit knowledge can then be
stored in knowledge bases or repositories for day-to-day use.
• From explicit to tacit knowledge is internalizing or fitting explicit knowledge into
tacit knowledge. An example is learning to ride a bicycle by practicing again and
again.
• From explicit to explicit knowledge is more or less combining, reorganizing,
categorizing, or sorting different bodies of explicit knowledge to lead to new
knowledge. An example of this would be sorting an employee master file to see how
many employees are older than 65 years of age, fall between the ages of 50 and 64,
and so forth.

11. What is a knowledge map? How does it differ from a decision tree? Be specific.

A knowledge map is a visual representation, not a repository of knowledge. It is a


straightforward directory that points to people, documents, and repositories. The main
purpose is to direct people where to go when they need certain expertise. In contrast, a
decision tree is composed of nodes representing goals and links that represent decisions
or outcomes.

12. Discuss the role of inferencing in knowledge-based codification.

Once the knowledge base is built, the system begins to make inferences, which means
derives a conclusion based on statements that only imply that conclusion. The inference
engine is the reasoning component that decides when certain inference rules should be
accessed or fired. Reasoning is the process of applying knowledge to arrive at solutions.

13. Why do we call software agents ‘agents’? Where do they fit in when it comes to
knowledge codification? Explain.

We call them agents, because like human agents, they represent human intelligence.
They can be programmed to interact with other agents (or humans) by some agent
communication language. An agent can also be programmed to learn from the user
behavior and deduce future behavior in order to assist the user.

KNOWLEDGE EXERCISES

1. Review the literature via the Internet and write a 2-page essay on CBR. What it is, how
it relates to knowledge codification, etc.

Dr. Faouzi Maddouri 7-3


CHAPTER 7
KNOWLEDGE CODIFICATION

Students should be encouraged to access Internet sites and focus on the latest on CBR to
benefit from recent literature or write ups.

2. Review the employment ads of two consecutive Sunday editions of a major newspaper.
What pattern in knowledge management jobs may be significant? What types of jobs
relate to knowledge management? What types of organizations are looking for
knowledge developers or specialists in the field?

The purpose of this exercise is to alert students to the availability of emerging jobs in
knowledge management. Very likely, advertised KM jobs are with large corporations
with ambitions to move forward into new products or new services. Of these companies,
most of them are likely in finance and banking, transportation, or food processing.

3. The knowledge developer’s work is to model the expertise, not the expert. Do you
agree with this sentence? Discuss its meaning.

Answers will vary.

4. Write a basic set of three rules showing how rules are constructed. Use at least one
AND and one OR operator in your rules.

The main focus of logic is:

IF course is KM_system AND instructor is Tremaine


THEN lectures are interesting

IF lectures are interesting or exam is fair


THEN class_size is large

IF class_size is large
THEN class_participation is low

5. If you were to explain to a novice the meaning and role of rapid prototyping, what you
would you say? Give an example to illustrate.

Rapid prototyping is a spontaneous, on-the-spot, iterative approach to building


knowledge automation systems; it is an iterative process by which the knowledge
developer shows the domain expert what the knowledge-based system will look like
based on the knowledge captured to date. This tool offers a number of benefits to
expediting the building of a knowledge base. For example, it demonstrates the technical
competence of the knowledge developer. It promotes innovation under the direct
guidance of the domain expert. It also reduces maintenance by up to 90 percent. One
thing obvious about rapid prototyping is the tendency to open up communication by
demonstrating rather than codifying the knowledge base. More than ever, the practice
reduces bugs by continuous testing and verification of each version. One should keep in

Dr. Faouzi Maddouri 7-4


CHAPTER 7
KNOWLEDGE CODIFICATION

mind, though, that unlike other tools, rapid prototyping lacks clear guidelines for use and
has a limited number of knowledge developers trained in its use.

6. Examine the following profiles of knowledge developers and discuss their


qualifications:
a. Erika and Lori were contracted to build a knowledge-base system for a country
club. The system is designed to improve the process of evaluating the applications
that the club receives for snack bar worker, lifeguard, and coach. Erika and Lori
hold B.S degrees in IT and are well versed in two knowledge software packages.
Erika is quite congenial and loves to talk to people. Lori is more introspective and
enjoys building knowledge systems without involvement with the user or the
company’s human experts. In addition, they are both well versed in Pascal and
C++. Neither has built a knowledge-based system for the club before.
b. Jeanine holds an MS in management information systems from a university in
Florida. She is well versed in expert systems and database design. Her favorite
database package is Oracle. She enjoys working with people and seems to thrive
under stress. She has developed several small knowledge bases. She tends to be
pushy when she needs information. People who worked with her in the past said
she is good at what she does but can be intimidating. She gives the impression she
has to be in command at all times, using others to do the “leg work.”
c. Bob is 26 years old and has an MS in computer science from a premier university
in New York. He served in the military for three years prior to going to graduate
school. Someone who worked with Bob on a recent knowledge-based project for
the military commented, “Bob has patience for complexity. He is a good listener
and is highly intelligent, but has a tendency to spin off on his own when the project
falls behind schedule, leaving the rest of his team behind.” He has trouble
delegating authority. He knows his technology inside out and feels he can do the
work on his own.
d. Ray earned a PhD in information technology from a Midwestern graduate school
in 1974. He has a master’s degree in social psychology from the same school.
Until the mid-1980s, Ray designed information systems in the banking sector. He
started working with software agents and knowledge-based systems in 1997 and
found it both challenging and lucrative. Recently, he worked on a medical
application with a renowned specialist. The two did not get along well. Ray
accused the expert of being uncooperative, and the expert told Ray he was
arrogant. The knowledge-based medical project ended in disaster.

Students should find this exercise challenging. In one respect, any of the four (a, b, c, d)
qualify with some coaching; yet, none of them are that ideal. Of the four candidates,
perhaps Erika and Lori are the closest to do the work. Here are some comments
regarding each knowledge developer:

a. Erika and Lori seem to get along well as a team, which is a prerequisite for team
work.They have the academic credentials and technical background (knowledge of
two shells, C++, Pascal) to do the job. Important too is Erika’s good nature and
communication skills and her ability to do a good job interfacing with a domain

Dr. Faouzi Maddouri 7-5


CHAPTER 7
KNOWLEDGE CODIFICATION

expert. Since the Hiring Advisor is not that complex and the nature of the problem
domain is not a “bread and butter” type of application for the club, Erika and Lori
could probably do a great job.

Based on the author’s experience, academically qualified people handling a basic


project for the first time tend to do well. In contrast, many experienced specialists
with a backlog of applications do not have the same commitment or devote prime
time for the effort.

The main question to consider in this case is whether it takes two knowledge
developers to build the knowledge-based system. This particular system was actually
built with 28 rules by Erika and Lori for two weeks. Erika worked at the club, which
helped in learning about the club, how it functions, its human resources needs, etc.
The other question is whether clients in the real world would want to hire
inexperienced consultants to do such a job. There is really no answer, other than to
pose it as a constraint.

b. As a single knowledge developer, Jeanine would be qualified to build the Hiring


Advisor. She is good with people and works well under stress, as dealing with
experts can be stressful. She already knows EXSYS Rulebook (a knowledge
automation development package) and probably can learn another shell with little
difficulty. Since the problem domain can be handled by one knowledge developer,
she can do the job.

The only constraint is her tendency to push her luck when she needs information.
Since all information comes from the expert, there is no telling what can happen.
Experts with egos or sensitivities can react adversely to arrogance or haughtiness.

c. Bob’s technical education makes him a natural for the knowledge codification (shell
programming) phase of building a knowledge-based system. Since the nature of the
problem domain can be handled by one knowledge developer, he would be qualified
to do the job. He does not have to be troubled by delegating work in a team setting or
worry about getting along with teammates. With his knowledge of EXSYS
Rulebook, he is probably the best qualified of the four candidates.

d. One possible constraint is Ray’s lack of experience in building knowledge-based


systems and his tendency to argue (in this case, with the expert) which might
backfire. Experience in information systems does not necessarily assure success in
building knowledge-based systems. The nature of information is different and the
thought process is captured differently in knowledge-based systems than information
systems. Capture tools can also be different. So, it is hard to predict how likely Ray
will work out. Clients in general look at a Ph.D with some respect, but most of them
also expect satisfactory results. We would give Ray a 50/50 chance, depending on
many factors such as the tolerance of the expert, the growing maturation of Ray, or
the combined interaction that helps both parties work together in building the
knowledge-based system.

Dr. Faouzi Maddouri 7-6


CHAPTER 7
KNOWLEDGE CODIFICATION

7. Someone suggested two types of potential knowledge developers: “Send me a well-


developed computer programmer or a programmer competent in several languages,
and we’ll make him or her into a successful knowledge developer.”

and

“Send me a talented generalist with well-developed interpersonal skills—or somewhat


more delicately, ‘a user friendly person’—and a rigorously analytical mind, and we’ll
team him or her with a competent knowledge developer.”

In your opinion, which approach would be more successful in knowledge


development? Why?

In building knowledge-based systems, communication and analytical skills are critical


requirements for working with this type of technology. There is hardly any algorithm
involved. We have also learned that someone with these skills can learn just as easily
how to write rules and test the resulting knowledge base as someone with programming
competency. So, the latter approach would be more attractive to successful knowledge
development than the former.

8. Develop a set of general frames to codify the following:


• A horse
• A student
• An airline pilot

Use these frames and describe the following:

• Flashdance, an 18-hand thoroughbred


• Brenda, a medium-height, fourth-year liberal arts students
• Fred, a 30-year veteran airline captain

Dr. Faouzi Maddouri 7-7


CHAPTER 7
KNOWLEDGE CODIFICATION

a.

Horse Frame

Families: Thoroughbred

Thoroughbred Frame
is a: animal
is a: 18 hand
jump: yes
trot: yes

Instance of
Mini: instance frame Facet Facet
is a: Thoroughbred Location: Location:
boards in: stall Stable Glen Farms
b.
College Frame

Families: student

Student Frame
is a: young human
is a: undergraduate
is a: equestrian

instance of
Mini: instance Frame
Name: Brenda
Height: medium Facet Facet
Major: liberal arts Location: dormitory Location: N.E. Quad
Year: first
Lives in: room

Dr. Faouzi Maddouri 7-8


CHAPTER 7
KNOWLEDGE CODIFICATION

c.
Airline Frame

Families: pilot

Pilot Frame
is a: human
is a: captain
is a: certified
is a: veteran

instance of
Mini: instance frame
is a: pilot Facet Facet
name: Fred Location: Chicago Location: Cook County
age: 30
lives in: suburb

Note that frames can be organized in different configurations. This is one such configuration.

Dr. Faouzi Maddouri 7-9

You might also like