Professional Documents
Culture Documents
2009
These notes are to supplement the textbook and provide other viewpoints. I apologize for
the lack of references. These notes were originally just lecture notes and not given out.
Part 1 Overview
Today’s Environment According to a recent survey, the top business
priorities, in order, from one to 10 were business-process improvement,
security breaches and disruptions, enterprise-wide operating costs,
supporting competitive advantage, data protection and privacy, the need for
revenue growth, using intelligence in products and services, focus on internal
controls, shortage of business skills and faster innovation and cycle times.
1
2. Programmable thermostats
The basics and fundamentals of the knowledge/DSS never talk about the decisions that
vary according to geographic locations. By geographic locations I mean, a problem in
decision in USA may not have the same solution for the same problem in India or else
where, because of geographic constraints. The constraints might be the people, their mind
sets and most important "buying power-Money".
Of course geographic locations affect decisions but so does culture, history, religion,
social status, education, financial, intelligence etc.
We are focusing on developing DSS techniques to make better decisions. Better decisions
must take all aspects under consideration. “Tailored approach”
Fundamental change
Small companies are often at the mercy of the larger corporations and their decisions
and they will be a part of their system and as a system they will never be able to make
decisions on their own. It’s all a chain mechanism. Example of UTA and
PeopleSoft, SAP
For example, in the 60’s only large companies had computers; today everyone does. In
fact at Lockheed Martin many other large companies use Excel. Excel does not cost
much and only requires a PC.
You could therefore ues Excel to schedule your labor pool, how to best buy power by
keeping records of energy costs, analyze them analytically and then make better
decisions, etc. Use excel to determine how much taxes you owe.
Look at the many low cost technologies that available today. Internet, cell phones with
internet, GPS and camera, smarter cars, ATM, Gas pumps with credit cards, bar codes,
RFID.
Many large companies transition to failure (Jeep, Digital Equipment, Enron, Pan Am,
locomotive companies, telegraph, etc. )
2
New large companies are being born every year – Google, Apple, Virgin Airlines,
Inofsys, Face Book
Some thoughts:
“It’s all about data, information and knowledge and knowing the difference.” J Priest,
2005.
Ordering pizza
Ordering A Pizza in 2010
This is scary because it might possibly be true one day. Be
sure that your speakers are on and the volume is turned up.
Just click the link below.
ORDERING PIZZA IN 2010
Need to differentiate - 1967 Webster’s dictionary
1. Data – factual material used as a basis for calculating or reasoning (factual may
not be correct!)
2. Information – Data that has been processed and presented in a form suitable for
human interpretation.
3. Knowledge – the fact or condition of knowing something with familiarity gained
through experience or association
4. Actionable knowledge – knowledge where correct or “best” decision can be
taken. (satisficing – efficient but not necessarily optimal)
Our focus
Knowledge (goal)- right collection of information at the right time (accurate, useful
and timely) or ( 1. contextual, 2. relevant, and 3. actionable) – lets go with blue
Knowledge (our task) - ability to detect relationships between objects and events,
and reason about those relationships to arrive at judgments, decisions and
3
conclusions. The ability to discern patterns and make sense out a sea of information
or lack there of (know when we do not know).
• KM Initiatives
o Knowledge acquisition/creation
o Knowledge Representation - Organize, format
o Application - Seeking / sourcing– pull Sharing – push
4
•
[edit] Approaches to knowledge management
Nonaka and Takeuchi (1995) argued that a successful KM program needs to, on the one
hand, convert internalized tacit knowledge into explicit codified knowledge in order to
share it, but also on the other hand for individuals and groups to internalize and make
personally meaningful codified knowledge once it is retrieved from the KM system.
Critics have however argued that Nonaka and Takeuchi's distinction between tacit and
explicit knowledge is oversimplified, and even that the notion of explicit knowledge is
self-contradictory.[1]
• 2 types:
o Explicit, leaky (policies, procedures, documented)
o Tacit , embedded, sticky, subjective, cognitive, and experiential learning
Role of IEs
IE skills aid knowledge management
-----------------------------------------------------------
-- IE knowledge a plus --
Knowledge management can be defined as the sum total of all activities
that enable the creation, storage, distribution, and application of
knowledge in organizations.
IEs can help plan for or even prevent redundancy and failures within an
organization by devising formulations that detail the fragility of an
organization's knowledge base.
5
These insights can be brought to bear on an organization's knowledge
transportation problem, for example. IEs can aid an organization by
studying the routing and movement of knowledge to determine the
efficient mechanisms for connecting people. IEs can also:
* Help resolve an organization's problems in many ways. Their knowledge
in the design of adaptive and self-repairing systems is pivotal.
* Aid organizations in devising appropriate measures that gauge the
quality of knowledge management systems.
* Help an organization in the difficult task of tacit knowledge
maintenance. Before an organization contemplates downsizing or mergers,
it should check how the new organization fits into the existing
knowledge map.
* Aid in the design of flexible systems that will enable growth and
evolution. Knowledge management systems in organizations should
accommodate modifications and updates that account for changes in an
organization.
The skills of IEs are valuable and salient for making knowledge
management a reality in organizations. For decades IEs have optimized,
industrialized, and engineered physical components of organizations.
The talents of industrial engineers can also be applied to the non-
physical components of an organization.
Harrahs Gambling
Data warehousing
Data mining, business analytics, BI
CRM customer relationship management
DSS decision support system
BAM Business Activity Monitoring
BPM – Business Process Management or business process modeling
SOAP – Simple Object Access Protocol, message based protocol based on XML for
assessing web
SOA – Service Oriented Architecture
6
• Efficiency and effectiveness (goals achieved vs. cost, inputs to outputs)
• Sub optimization conflicting goals
• Satisficing often your goal is to identify several options for management or
you to decide on, 3 to 5 alternatives is the best , not too many, not too few
• Risk and scenarios risk vs, sense of control scenarios are perfect for IEs, think of
time studies where the focus is on the decision making and tasks performed not
the time, worst, best most likely, average way to handle chaos
Model Types
• strategic (top mgt, long term. Planning),
• tactical (how to make it happen, middle mgt, allocate and control resources),
operational (day to day, all levels),
• analytical (analysis of data, often specialists)
Other issues
• User interface – the future, text to speech, speech, 3d visual, animation, cell
phones PDA, RFID, wireless, UWB capabilities
• Blackboard
• Emerging use of spreadsheets (learn to use) this may be your tool
• Fedex web focus , case application overall discussion how things have changed.
Overnight to tracking in real time of trucks
• Static vs. Dynamic future focus will be on dynamic since we will soon be getting
real time data then information
Book Outline
An Introduction to Knowledge Engineering
Kendal and Creen
1. Introduction
a. Definitions - handout
2. Types of KBS
a. Expert systems
b. Neural
c. CBR
d. Genetic
e. Intelligent agents
f. Data mining (databases)
3. Knowledge acquisition
a. Explicit, tacit, Deep, shallow
b. Unstructured
c. Structured
d. Event recall
e. Twenty questions
f. Added observation, questionnaire, scenario, storyboard
4. Knowledge representation and reasoning
7
a. Heuristics
b. Propositional logic and symbols
c. Forward and backward chaining, Data driven, goal driven
d. Problems of explanation and brittleness
e. Semantic networks
f. Frames
g. Ontology - handout
5. Expert shells
a. What are shells
b. AI languages, PROLOG, Facts, inferences
6. Life cycles and methodology
a. Product development
b. Importance of prototypes
c. Blackboard architecture
d. Problem solving
e. KADS
f. Hybrid
g. AION BRE
7. Uncertain reasoning
a. Handout
b. Confidence factors
c. Probabilistic reasoning, Bayesian
d. Fuzzy Logic - handout
8. Hybrids
Key questions
• When are these methods used?
• Advantages and disadvantages
• Concerns
Chapter 2 F Databases
8
2. Relational database – DBMS that manages data as a collection of tables in
which all data relationships are represented by common values (ID #) in
related tables
3. Object oriented database
4. Entity – thing about which an organization chooses to record
5. Attribute – Named property or characteristic of an entity that is of interest to
the organization
6. Object – Structure that encapsulates attributes and methods that operate on
those attributes
7. Object class - Logical grouping of objects that have the same or similar
attributes and behavior
8. Primary key – A candidate key that has been selected as identifier for an entity
type, may not be null
9. Candidate key – An attribute that uniquely identifies each instance of an entity
10. Business rule – Specifications that preserve the integrity of the logical data
model
11. Trigger – Assertion or rule that governs the validity of data manipulation
operations such as insert, update, and delete
12. SQL – fourth generation query language
9
• Larger companies more likely
• Successful companies spend 50% more on BI, necessary but not sufficient for
success
• Dissatisfaction usually derives from distribution of the results
Data Mining
Data mining – Most important, can be IE job
2. Used to describe knowledge discovery in databases
3. Fancy words for data analysis not just statistical, find new
knowledge/information
4. Key word search, lack of depth, Google itus
5. Statistics
6. 3 types of methods (simple statistics, intermediate (regression, decision
trees, and clustering) and complex (neural, rule induction, genetic, CBR)
7. Predictive or descriptive
8. Uses
1. Typical – predictive, who is buying what, loans, explain what is
happening at a store, etc.
2. Multiple decisions – decision trees
3. Early warning (anomaly detection) – identify problems quickly, notify
right people, respond automatically or manually, identify emerging
threats, sudden increases, certain conditions, tire recall, keeping track
of patent disclosures
4. Survival data mining – time to event problem, came from medical
studies of dying patients, churn, migration, identifying lapsed
customers, “when” will a customer make a new purchase or upgrade
(10% within one year, 30% one to two year, etc.), customer half life
9. How organizations grow
i. What happened - reporting
ii. Why did it happen - analysis
iii. What will happen – prediction
iv. What is happening - operations
v. What do I want to happen – active BI
10. Predictive analysis to avoid traffic jams – GPS and radio traffic updates
11. Recognizing customers and what they want before they enter a fast food
restaurant, monitor # of cars entering, 50% people at lunch order
cheeseburger, avg. 2 people per car, gives chance to prepare food quicker
12. Tools, list notice rehash
13. Text mining major area for future
14. Why training, validation, verification
11. Clustering – analyze historical data and automatically generate a model that cab
predict future behavior. Partition a database into segments.
Most effective when number of variables involved is very large and
the relationships between them is complex and imprecise
10
Neural networks – difficult to provide good rationale for the
predictions. Need considerable training data. Limits to size of
database (large and small)
Types
1. Nearest neighbor kNN, LRW
2. Neural network
3. Rule induction
4. Decision trees
5. Time series
12. New territory
1. Unstructured and semi-structured data (web content and real time)
2.
13. Data visualization future area
1. Haptics, action virtual touch
14. Multidimensional dimensions, measures, and time
15. Real time data collection and analytics, and control/action (Autonomous)
16. Web intelligence BI and the web
17. Case study, Cluster Analysis handout
Knowledge elicitation – obtaining knowledge from a human expert (e.g. domain expert)
(or a database)
11
o They don't know all that they know and use
12
• Heuristics - "rules of thumb" used to make a decision.
13
• Conduct an initial interview with the expert in order to (a) scope what knowledge is to
be acquired, (b) determine what purpose the knowledge is to be put, (c) gain some
understanding of key terminology, and (d) build a rapport with the expert. This
interview (as with all session with experts) is recorded on either audiotape or
videotape.
• Transcribe the initial interview and analyze the resulting protocol. Create a concept
ladder of the resulting knowledge to provide a broad representation of the knowledge
in the domain. Use the ladder to produce a set of questions which cover the essential
issues across the domain and which serve the goals of the knowledge acquisition
project.
• Conduct a semi-structured interview with the expert using the pre-prepared questions
to provide structure and focus.
• Transcribe the semi-structured interview and analyze the resulting protocol for the
knowledge types present. Typically these would be concepts, attributes, values,
relationships, tasks and rules.
• Represent these knowledge elements using the most appropriate knowledge models,
e.g. ladders, grids, network diagrams, hypertext, etc. In addition, document
anecdotes, illustrations and explanations in a structured manner using hypertext and
template headings.
• Use the resulting knowledge models and structured text with contrived techniques
such as laddering, think aloud problem-solving, twenty questions and repertory grid to
allow the expert to modify and expand on the knowledge already captured.
• Repeat the analysis, model building and acquisition sessions until the expert and
knowledge engineer are happy that the goals of the project have been realized.
• Validate the knowledge acquired with other experts, and make modifications where
necessary.
Techniques have been developed to assist this, such as the use of ontologies and problem-
solving models. These provide generic knowledge to suggest ideas to the expert such as
general classes of objects in the domain and general ways in which tasks are performed.
This re-use of knowledge is the essence of making the knowledge acquisition process as
efficient and effective as possible.
14
Unstructured – best for innovation and very complex?
Structured – most common, table 11.1 procedures
Process tracking or protocol analysis – cognitive method
Observations (time studies)
Case analysis or SEP– often best most organized
Critical incident
Users rather than experts (Do you want to know the time or how to
make a watch)
Brainstorming
Prototyping
o Semi-automatic
RGA
ETS (expertise transfer system)
o Automatic
Machine learning
Inductive learning
Neural networks
Genetic algorithms
• Multiple experts vs. one expert
o Almost have to use scenarios or cases
• Automated rule induction
o Induction is the process of reasoning from the specific to the general
o For us, where a computer generates rules from example cases
• Advantages of rule induction
o Certain, small, loosely coupled, or modular
o Exponential complexity
o Difficulties- too long a list
• Knowledge verification and validation
o Critical for success
• Representation of knowledge (very important)
o Ontology and taxonomy – words are important, handouts!!!
o Production rules
Declarative rules
Inference rules
o Semantic networks
Relationships between different concepts
Inheritance
o Frames and Objects
Heirarchy and inheritance
o Object oriented representation,
o Decision tables - popular
o Decision trees – popular
15
o Predicate calculus or predicate logic or propositional logic or first order
logic
• Reasoning
• Forward and backward chaining
o Examples are very important, best method to explain concept
o Remember multi-level rule
o Real world problems in rule firing and process – handout
• Inference tree
• Explanation – important for professionals or when other factors outside of the
system can or should effect decision
16
b. Features
c. Structure
• How expert systems work
a. Rules
b. Uncertainty processing
c. Inference engine
d. Forward and backward
Collaborative
(this is a fact of life because of expertise of each discipline, but some will not) lets all
communicate better
1. groupwork, discuss role of portals
2. Groupware, brainstorming hard to do over internet
3. Group support systems technologies (Lotus Notes)
4. No distance learning, but discuss
5. Portal market,
6. Supply chain, RFID, GPS auto
7. Problems in supply chain, uncertainty, demand forecast uncertainity so what do
you do?
8. Nikes famous problem, i2,
9. Future EIS systems, 1. real time, 2. easier to use, 3. adaptable to each exec.
IS over Internet
17
1. Service-oriented architecture, SOA, simply stated, is a computing architecture
where application functionality is available as shared services on a network.
Services are what we know as Web services
E-Commerce
Agents
5. Agent – computer programs that simulate a human relationship by doing
something that another person could do for you (Selker), behave in a manner like
human agents e.g. travel agents, insurance agents. Can be classified by
a. Autonomous, software, intelligent, interface, virtual, informational,
mobile, etc.
b. Applications such as imaging, business process, etc.
c. Notions
i. Weak notion of agency – involves autonomy or the ability to
function without intervention (Wooldridge and Jennings)
ii. Strong notion of agency – uses mental components such as belief,
desire, intention, knowledge, etc.
6. Multi-agent systems – typically distributed systems in which several distinct
components, each of which is an independent problem-solving agent come
together to form some coherent whole. For multi-agents systems must consider
agent modeling of its world, multi-agent planning to coordinate their behavior,
social relationships with other agents (cooperation),interactions and
communication
7. SMART Agent Framework (d’Inverno and Luck, 2001)
a. Entity and environment – simple collection of attributes
b. Object – also a collection of attributes but a more detailed description may
be given and in addition describing their capabilities. Capabilities of an
object are defined by a set of action primitives which can be performed by
the object in some environment and consequently change the
environment’s state.
c. Agent – an object with goals
d. Autonomous agent – agent with motivation, self-motivated agents in the
sense they can create and pursue their agendas as opposed to functioning
under the control of another agent. Goals generated from motivations
within not from others.
8. Primitives of agent models
a. Attributes – perceivable feature, simple features of the world, tree is green,
in a park, 20 feet high
b. Actions – discrete event that can change the state of the environment when
performed, attach tires to a car, get tire, move from one wheel to the next,
delete the attribute that the robot is at the first wheel, add the attribute that
agent is at the second.
c. Goal – state of affairs to be achieved in the environment, robot’s goal is to
attach tire to the wheel.
18
d. Motivation – any desire or preference that can lead to the generation and
adoption of goals and that affects the outcome of reasoning or behavioral
task intended to satisfy those goals
Life is full of uncertainty. Even the universe and math is full of uncertainty.
Source: Intelligent Systems for Engineers; Adrian Hopgood, CRC Press, 2001
Overall there are situations in nature that are truly random. Some can be
anticipated on a statistical basis. For example the roll of a dice has a 1/6 change
of being a 2.
19
Exceptions and uncertainty are different.
• Uncertain means not certain to occur, unreliable.
• Exception is a case to which a rule does not apply. An occurrence
that is not normal and usually unexpected.
• Random – elements with probability of occurrence
One way to think about is that uncertainty (the unknown future) causes exceptions
(actual situations not normal and/or not seen before).
Representing uncertainty
• Numeric - %, many prefer ranking, influence diagrams
• Symbolic – Likert scale
1. Influence diagrams look at figure
20
2. Spreadsheets
3. Decision tables and trees **
4. LP and more OR you should know this
5. Simulation, especially when to use
6. Visual interactive randomness in video games
21