KNOWLEDGE

CODIFICATION
What Does Knowledge
Codification Involve?
 Converting “tacit knowledge” into
“explicit usable form”
 Converting “undocumented” information
into “documented” information
 Representing and organizing knowledge
before it is accessed
 It is making institutional knowledge
visible, accessible, and usable for
decision making
 Benefits of Knowledge
Codification
 Instruction/training—promoting training of
junior personnel based on captured
knowledge of senior employees
 Prediction—inferring the likely outcome of a
given situation and flashing a proper warning
or suggestion for corrective action
 Diagnosis—addressing identifiable symptoms
of specific causal factors
 Planning/scheduling—mapping out an entire
course of action before any steps are taken
 Pre-KC Questions
 What organizational
goals will the codified
knowledge serve?
 Why is the knowledge
useful?
 Howwould one codify
knowledge?
Some Codification Tools
 Knowledge Map
 Decision Table
 Decision Tree
 Frames
 Production Rules
 Case-based Reasoning
Knowledge Map
 Visual representation of knowledge, not a
repository
 Identify strengths to exploit and missing
knowledge gaps to fill
 Can be applied in Knowledge Capture
 A straightforward directory that points people
to where they can find certain expertise
 Capture both explicit and tacit knowledge in
documents and in experts’ heads
 Knowledge Map (Relationships
among Departments)

www.nwlnk.com Copyright 2004

 The Building Cycle
 Once where knowledge
resides is known, simply
point to it and add
instructions on how to get
there
 An intranet is a common
medium for publishing
knowledge maps
 Main criteria: clarity of
purpose, ease of use,
accuracy of content
Decision Trees
 Composed of nodes representing goals and
links representing decisions or outcomes
 All nodes except the root node are instances
of the primary goal. (See next figure)
 Often a step before actual codification
 Ability to verify logic graphically in problems
involving complex situations that result in a
limited number of actions
Discount Policy (A Decision Tree)
Discount ?
Order 6 or Discount
size ? more is 25%
Customer is copies
bookstore
Discount ? Discount
Less
is NIL
than 6
copies
Bookstore

Discount Discount ?
50 or Discount
Policy is 15%
more
copies
Not a
bookstore Discount ?
Order Discount
20-49
size ? copies is 10%
Customer is
library or
Discount ?
individual 6-19 Discount
copies is 5%

Less than Discount ? Discount

6 copies is NIL

6-10
Decision Tables
 More like a spreadsheet—divided into a
list of conditions and their respective
values and a list of conclusions
 Conditions are matched against
conclusions (See next table)
Discount Policy (A Decision Table)
Condition Stub Condition Entry
1 2 3 4 5 6

Customer is bookstore Y Y N N N N

Order size > 6 copies Y N N N N N

Customer is librarian/individual Y Y Y Y
IF Order size 50 copies or more Y N N N
(condition) Order size 20-49 copies Y N N
Order size 6-19 copies Y N

Allow 25% discount X

Allow 15% discount X
Allow 10% discount X
THEN Allow 5% discount X
(action) Allow no discount X X

Action Stub Action Entry

Frames
 Represent knowledge about a particular idea
in a data structure
 Handle a combination of declarative and
operational knowledge, which make it easier
to understand the problem domain
 A frame represents knowledge about an
entity in the real world, such as an employee,
a person.

6-13
Frames
 Have a slot (a specific object or an attribute of
an entity) and a facet (the value of an object
or a slot)
 When all the slots are filled with values, the
frame is considered instantiated which
means instance of the frame is created.

7-14
 Range: set of possible values
 Default (value to assume if none were
specified.

6-15
 Generic COUPE Frame
An Automobile
Specialization:
Example AUTOMOBILE
Generalization:
Generic AUTOMOBILE
(SMITH’S AUTOMOBILE,
Frame
HANSON’S AUTOMOBILE)
Specialization:
Doors: 2
VEHICLE
Generalization:
(STATION-WAGON,
COUPE, SEDAN) SMITH’S AUTOMOBILE
. Frame
. Specialization:
. COUPE
Year: .
Range: (1940 – 1990) .
If-Changed: (ERROR: .
Value cannot be modified)
Year: 1990
.
. Doors: ( )
.
Production Rules
 Tacit knowledge codification in the form of
premise-action pairs
 Rules are conditional statement that specify an
action to be taken if a certain condition is true
 The form is IF… THEN, or IF…THEN…ELSE
 Example:
IF income is “average” and pay_history is “good”
THEN recommendation is “approve loan”
6-17
Case-Based Reasoning
(CBR)
 CBR is reasoning from relevant past cases in
a manner similar to humans’ use of past
experiences to arrive at conclusions
 Goal is to bring up the most similar historical
cases that match the current case
 Requires rigorous initial planning of all
possible variables
 Referringto old cases is advantageous
when dealing with recurring situations.
Because no old case is exactly same, old
solutions must be adapted to fit it.
Generic CBR Process
Submits Partial Description
Specify Attributes of
User of a New
Problem
Problem

Similar
Cases Match Attributes
Case Base to Those in
Case Base

User
Role of Planning
In knowledge-based systems, planning
involves:
 Breaking the KM system into module
 Looking at partial solutions
 Linking partial solutions via rules and
procedures to arrive at final solutions
 Deciding on the programming language
 Selecting the right software package
Role of Planning (cont’d)
 Arranging for the verification and
validation of the system
 Developing user interface and
consultation facilities
 Promoting clarity and flexibility
 Reducing unnecessary risks
 Making rules easier to review and
understand
Knowlwdge Deveoper skill set
1. Knowledge Requirements:
(a) Computer Technology: well versed in
software, hardware, operating system,
security, testing, documentation and
maintenance
(b) (b) domain specific knowledge: familiar
with nature of problem, business of
user organization political climate, level
of management support
© Knowledge repositories and Data
mining: familiarity with knowledge
repositories, and what off the shelf
packages available for data
warehousing and data mining
(d)Cognitive Psychology : understanding
knowledge in general and how experts
think
2. Skill Requirements:
(a) Interpersonal communication
(b) Ability to articulate project’s rationale:
project feasibility, justifying investment
(c) Rapid prototyping skills: sit down with
expert to get feedback on how the
knowledge base is shaping up.
(d) Personality attributes: Intelligence,
creativity, Tolerance, Realism,
Persistence, Optimism