CPE/CSC 580:
Knowledge Management
Dr. Franz J. Kurfess
Computer Science Department
Cal Poly
© 2001 Franz J. Kurfess
Knowledge Presentation 1
Course Overview
 Introduction
 Knowledge

Knowledge Acquisition,
Representation and
Manipulation
 Knowledge




Retrieval
Information Retrieval
Knowledge Navigation
 Knowledge


Presentation
Exchange
Knowledge Capture, Transfer,
and Distribution
 Usage
Organization
Classification, Categorization
Ontologies, Taxonomies,
Thesauri
 Knowledge

Processing
 Knowledge
of Knowledge
Access Patterns, User
Feedback
 Knowledge
Management
Techniques

Topic Maps, Agents
 Knowledge
Management
Tools
 Knowledge Management in
Organizations
Knowledge Visualization
© 2001 Franz J. Kurfess
Knowledge Presentation 2
Overview Knowledge Presentation
 Motivation
 Alternatives
 Objectives

 Evaluation

Criteria
 Chapter Introduction

Cognition and Perception
 Visualization



Data Visualization
Information Visualization
Knowledge Visualization
© 2001 Franz J. Kurfess
Sound
Tactile Presentation
 Virtual

to Visualization
Reality
Immersion
 Important
Concepts and
Terms
 Chapter Summary
Knowledge Presentation 3
Logistics
 Introductions
 Course
Materials
 textbook
 handouts
 Web
page
 CourseInfo/Blackboard System and Alternatives
 Term
Project
 Lab and Homework Assignments
 Exams
 Grading
© 2001 Franz J. Kurfess
Knowledge Presentation 4
Stories,
not Bories!

lengthy explanations


endless bulleted lists
allows the discussion of more
or less relevant issues at
arbitrary levels of detail


speaker reads from the
display


important stuff in unreadably
small font sizes
reinforces knowledge transfer
by exploiting multiple
modalities
audience enjoys a nice nap


no nasty/embarrassing
questions
refreshed after the talk
© 2001 Franz J. Kurfess
Knowledge Presentation 5
[Clemens 1998, http://www.idiagram.com]
Motivation
 reduce
the user’s overhead for locating and
interpreting knowledge
 indicate
attributes of concepts
 show relationships between items
 display results of a search
 visual
computing
 utilize

instead of “paper replication”
 make

the features of computer displays
technology conform to people
instead of forcing people to adapt to technology
© 2001 Franz J. Kurfess
Knowledge Presentation 7
Objectives
 be
aware of perceptual and cognitive aspects of
human information processing
 main
emphasis on visual input
 understand
presentation methods and techniques for
data, information, and knowledge
 evaluate the use of visualization techniques for
knowledge management purposes
© 2001 Franz J. Kurfess
Knowledge Presentation 8
Cognition
 cognitive
engineering
 design
principles for presentation techniques
 based on cognitive processes in humans

information processing, attention, memory
 main
emphasis on the visual system
 mental depiction can be as important as mental description
© 2001 Franz J. Kurfess
[Kowalski 1997]
Knowledge Presentation 10
Perception
 interface
between our mind and the world
 sensory information translates physical aspects of
the world into neural encodings in our brain
 visual
and auditory systems are most relevant for
knowledge-related perception
 many lower-level processing steps are encoded in
“wetware” and happen sub-consciously
© 2001 Franz J. Kurfess
[Kowalski 1997]
Knowledge Presentation 11
Information Visualization
 utilizes
the visual system to indicate important
aspects of data and information
 absence/presence
 quantity
 features
 basis
for writing, drawing, art
 long-distance
communication
 long-term preservation of knowledge
 graphical
displays offer a much richer visual
experience than text-based terminals
 flexibility,
© 2001 Franz J. Kurfess
resolution, color
[Kowalski 1997]
Knowledge Presentation 12
Cognitive Aspects of Vision
 proximity
 nearby
items are grouped together
 similarity
 similar
items are grouped together
 continuity
 smooth
continuous patterns vs. separate items
 closure
 automatic
filling of gaps in a figure
 connectedness
 interpretation
of related items as single units
many of these aspects are performed at low levels of perception
© 2001 Franz J. Kurfess
[Kowalski 1997]
Knowledge Presentation 13
Visualization Primitives
built-in, low level functions of our visual system
 orientation of shapes
 easy
detection of groupings
 color
 preference
for primary colors
 depth
 cues
to size, distance of objects
 arrangement
 deviation
 spatial
of objects
from regular arrangements are easily detected
frequency
 construction
© 2001 Franz J. Kurfess
of a coherent visual image is attempted
[Kowalski 1997]
Knowledge Presentation 14
Technology: Visual Computing
 computer
presentation technology has some
advantages over other media
 modify

representations of data and information
e.g. change color, scale
 show
changes in space and time through animation
 use interaction with the user to optimize presentation

according to the user’s preferences
 show

relationships between items
e.g. through hyperlinks
© 2001 Franz J. Kurfess
Knowledge Presentation 15
Visual Presentation Techniques
 text
 mostly
sequential
 good for details, explanations
 diagrams
 two-dimensional
 good
for structural aspects, relations between items,
properties
 images
 two-dimensional
 (partial)
reproduction of real-world objects
 creation of imaginary objects
 J.
e.g.
art
© 2001 Franz
Kurfess
Knowledge Presentation 16
Visual Presentation Methods
 hierarchical
structures (trees)
 appropriate

for items with relations such as
is-a, part-of, parent-child, dependencies, etc.
 becomes
difficult to use for large structures
 map
 arranges

items according to spatial proximity
useful for properties that map into space
 with
zooming, it can be used for large sets of items
 grid
 visualization

of tabular data
requires strong regularities in the overall information space
© 2001 Franz J. Kurfess
[Kowalski 1997]
Knowledge Presentation 17
Visual Presentation Methods cont.
 network
 items
(graph)
are represented as nodes, and relationships as arcs
 clusters
 related
 bar
items are grouped together
chart
 indicates
values of properties
 histogram
 shows
the distribution of items
 perspective
wall
 main
focus on the centerpiece (front), with less relevant
items arranged on the side panels
© 2001 Franz J. Kurfess
[Kowalski 1997]
Knowledge Presentation 18
Auditory Presentation Techniques
 language
 sequential
 similar
to text
 sound
 (partial)
reproduction of real-world events
 creation of new events

e.g. music
© 2001 Franz J. Kurfess
Knowledge Presentation 19
Relevance of Knowledge
Presentation
 better
user experience
 shorter
time to locate and identify relevant knowledge
 knowledge is easier to comprehend and utilize
 improved
understanding
 critical
examination of existing bodies of knowledge
 exploration and validation of relationships
 suitable presentation of abstract concepts
 creation
of new knowledge
 integration
of existing diverse bodies of knowledge
 addition of relationships between knowledge items
© 2001 Franz J. Kurfess
Knowledge Presentation 20
Data Visualization
 visual
display of data values
25
20
15
10
5
0
© 2001 Franz J. Kurfess
3-D Column 1
3-D Column 2
0
50
54
58
62
66
70
74
78
82
86
90
3-D Column 15
Knowledge Presentation 21
Information Visualization
 display
 e.g.
of relationships for structured data
entity-relationship diagrams
 document
clustering
 present
the user with a visual representation of the
document space constrained by the search criteria
 group related documents together

requires a similarity measure
 search
formulation analysis
 display
the relationships between various aspects of the
search terms and the retrieved results

effects of expansion, relevance feedback, etc.
 used
to help the user formulate a better query
© 2001 Franz J. Kurfess
Knowledge Presentation 22
Knowledge Visualization
 link
display
 indicates
relationships between items
 color, patterns, thickness, arrows, labels, etc. can be used
to differentiate types of relationships
 link
analysis
 correlates
multiple documents that share certain aspects
 helps with the identification of dependencies, trends, etc.
© 2001 Franz J. Kurfess
Knowledge Presentation 23
Alternatives to Visualization
 utilization
of other senses for the presentation of
knowledge
 auditory
speech
 signals


beeps
 tactile

virtual reality
 taste
 smell
© 2001 Franz J. Kurfess
Knowledge Presentation 24
Sound
 speech
 somewhat
limited due to the sequential nature
 helpful as alternative or additional method
 sounds
 sometimes
used for alerts, or to augment aspects of visual
display
 music
 primarily
used for entertainment purposes
 may be used to evoke emotional responses
© 2001 Franz J. Kurfess
Knowledge Presentation 25
Tactile Presentation
 Braille
 as
alternative to text input for visually impaired people
 virtual
reality
 mainly
augmentation of visual input
 special-purpose
 feedback


mouse
special mouse/mouse pad combination that delivers some tactile
feedback to the user
 feedback

devices
joysticks, haptic gloves
force feedback
used for tele-manipulation, VR
© 2001 Franz J. Kurfess
Knowledge Presentation 26
Virtual Reality
 tries
to provide a computer-based model of an
environment
 relies mainly on 3D visual input
 feedback between user and system is critical
 direct
manipulation of virtual objects
 mostly
used for modeling purposes, not so much for
knowledge presentation
© 2001 Franz J. Kurfess
Knowledge Presentation 27
Immersion
 similar
to VR, tele-presence
 the user has the impression of being in another
environment
© 2001 Franz J. Kurfess
Knowledge Presentation 28
Inxight
Tree




tree displays the
hierarchical
structure of a
Web site
overview of
available
contents
quick navigation
no details
[Inxight 2001]
© 2001 Franz J. Kurfess
Knowledge Presentation 29
Lexis-Nexis Tree
 built
with Inxight
Tree Studio
© 2001 Franz J. Kurfess
Knowledge Presentation 30
Understanding USA: Environment
© 2001 Franz J. Kurfess
[Understanding USA]
Knowledge Presentation 32
ClearForest ClearSight Viewer
[ClearForest 2001]
© 2001 Franz J. Kurfess
Knowledge Presentation 33
Outline
 Definitions
 Constant
Information Density
 VISAGE
 Big
Issues
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 34
Definitions
 Scientific
visualization
 Information visualization
 Database visualization
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 35
Definitions
 Scientific
visualization
 Physical
data, physical processes
 Information
 Abstract
 Database
visualization
representations
visualization
 Mapping
data in database
to graphical display
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 36
Outline
 Definitions
 Constant
Information Density
 VISAGE
 Big
Issues
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 37
Motivation
 Clutter
can have negative effects
 Decreased
user performance
 Diminished visual appeal
…
 VIDEO
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 38
Comments?
 Do
people buy the concept of Constant Information
Density?
 What tasks might it be appropriate for?
 Will a new type of graphic replace existing
paradigms?
 Playfair...
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 39
View from a high elevation
- This visualization has
three layers.
- At the user’s current
At the shown
user’s current
elevation,
by the
elevation,
only
thethe
horizontal line, only
outline
layer
statestate
outline
layer
is is
visible
visible.
We can see the
objects associated with
this layer in the canvas
on the left.
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 40
View from an intermediate elevation
The cities circles layer
becomes visible when
the user zooms
States are still
visible and a new
layer is visible
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 41
View from a low elevation
At this elevation, a
The layer
graphislayer
new
visible,
replaces
the layer
and
the circles
layer visible,
iscircles
no longer
so a graph of
transportation data
for each city replaces
the circle for each city
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 42
Width bars in our example
The two layers on the left contain
political boundaries. The four layers
on the right contain cities of different
sizes. We can see that there are
different numbers of cities in each of
these layers.
A couple of slides ago we saw a cluttered
visualization. This picture shows VIDA’s
density feedback on that same
visualization, using # objects as the
density function.
We can see that at
this particular
elevation, this layer
on the far right is
causing us the
trouble. We can
also see that other
elevations will be
cluttered.
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 43
After the user has made adjustments
User can make all the
necessary adjustments
without zooming back and
forth to check each change
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 44
Housing cost/Income visualization
VIDA chooses between these two
representations based on density as measured
by the # vertices - in the denser regions, dots
are drawn. In less dense regions, the state
outlines are drawn. Essentially, the user is
shown more detail about outliers.
Before
- This is a visualization of
housing cost (shown on the x
axis) versus income (shown on
the y axis).
- States can be represented as
either
dot or
a state outline.
© 2001aFranz
J. Kurfess
[Woodruff 1998]
After
Knowledge Presentation 45
Fortune 500 visualization
 Displayed
using VIDA’s
technique for non-uniform
data
Outliers have more detailed
representation
# employees

I’ve extended VIDA to support this
functionality as I’ll illustrate in this example of
Fortune 500 data
© 2001 Franz J. Kurfess
[Woodruff 1998]
% profit
Knowledge Presentation 46
Population visualization
Once the user has expressed their
constraints, VIDA can choose different
combinations of layers to show in
different parts of the visualization
Smaller cities are omitted in the more
populous areas in which larger cities are
drawn. In the less populous areas where no
larger cities exist, the smaller cities are
drawn.
Before
© 2001 Franz J. Kurfess
After
[Woodruff 1998]
Knowledge Presentation 47
Comprehensive list of actions
that can be performed on the
contents of a single layer to
decrease data density
Not every action affects every density function
Actions may affect several density functions
Operations to decrease density
<<graphical operations>>
Zoomed-in view of
circles showing
populations Original
of cities;
these are Baltimore
and Washington, D.C.
Change shape
Decreases amount
of ink
<<Data operations>>
Reduce size
Select
For example,
Decreases # objects
Decreases amount
of ink
Remove attribute
association
Decreases
data
Aggregate
Aggregate cities by state;
Chesapeake Bay;
Decreases # objects
density
Change color
Reclassify
Have only two sizes
of cities;
Decreases # sizes
© 2001 Franz J. Kurfess
[Woodruff 1998]
Notice how visually
different all these
options are
Decreases # colors
once we take other
layers into account
Knowledge Presentation 48
- Portals are a natural
mechanism for presenting a
number of visualizations to the
user.
<<This visualization in the upperright is particularly interesting.
The association between city
circle size and population has
been removed, and an
association between state color
and population has been
added.>>
The transformation canvas
 Transformations
presented
to user in a “transformation”
canvas

Each transformation appears
as a portal
 When
the user zooms, the
visualizations all change
 How should user navigate
Portals’ contents
the transformation
space?
change dynamically
© 2001 Franz J. Kurfess
-- VIDA should give transformations like the one the user selected
- After user enters the portal for a transformation, they can edit it
- VIDA could then show them transformations that incorporated
their changes [Woodruff 1998]
Knowledge Presentation 49
Outline
 Definitions
 Constant
Information Density
 VISAGE
 Big
Issues
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 50
Issues
 What
are some of the good techniques?
 What are some of the problems with VISAGE from
the user’s perspective?
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 51
Outline
 Definitions
 Constant
Information Density
 VISAGE
 Big
Issues
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 52
Interesting things to visualize
 Graphs
 The
Web!!!
 Networks
 Network
 Financial
 Data
administration
data
mining (very little visual data mining work to date)
 etc...
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 53
Open problems in visualization
 What
are the limits of visualization?
 Multi-dimensional analysis
 Context, context, context
 Users
get lost while navigating
 Users lose linkages in multiple views
 Users get confused composing Magic Lenses
 Creation
 End-user
programming, automated creation
 Misrepresentation
© 2001 Franz J. Kurfess
[Woodruff 1998]
Knowledge Presentation 54
Knowledge Trails
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 55
Knowledge Levels in Organizations
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 56
Knowledge Landscapes
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 57
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 58
© 2001 Franz J. Kurfess
Knowledge Presentation 59
Visualization at NorthernLight
© 2001 Franz J. Kurfess
Knowledge Presentation 60
Scatter Plots
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 61
Pie and Bar Charts
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 62
Turkey Box Plot
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 63
Multidimensional Data
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 64
DBMiner
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 65
DBMiner 2
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 66
Daisy Chart
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 67
Clementine
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 68
Decision Tree Visualization 1
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 69
Decision Tree Visualization 2
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 70
Decision Tree Visualization 3
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 71
Scientific Visualization
© 2001 Franz J. Kurfess
[Cook 1999]
Knowledge Presentation 72
Human Effectiveness
© 2001 Franz J. Kurfess
[http://www.idiagram.com/]
Knowledge Presentation 73
Semiotic Model
[http://www.idiagram.com/]
© 2001 Franz J. Kurfess
Knowledge Presentation 74
Semiotic Triangle
[http://www.idiagram.com/]
© 2001 Franz J. Kurfess
Knowledge Presentation 75
Knowledge Visualization
[Clemens 1998, http://www.idiagram.com/kv_venn.html]
© 2001 Franz J. Kurfess
Knowledge Presentation 76
Characteristics of Complex Systems
[Clemens
© 2001 1998,
Franzhttp://www.idiagram.com]
J. Kurfess
Knowledge Presentation 77
System Representation
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 78
System Dynamics
© 2001 Franz
J. Kurfess
[Clemens
1998, http://www.idiagram.com]
Knowledge Presentation 79
System Hierarchies
© 2001 Franz J. Kurfess
Knowledge Presentation 80
[Clemens 1998, http://www.idiagram.com]
Complex Adaptive Systems Model
[Clemens 1998, http://www.idiagram.com]
© 2001 Franz J. Kurfess
Knowledge Presentation 81
Evolutionary System Model
[Clemens 1998, http://www.idiagram.com]
© 2001 Franz J. Kurfess
Knowledge Presentation 82
Visualizing Complex Systems
© 2001 Franz J. Kurfess
[Clemens 1998, http://www.idiagram.com]
Knowledge Presentation 83
References
[Cook 1999] Diane Cook: Data Visualization, Course at UT Arlington, 1999.
http://ranger.uta.edu/~cook/dm/lectures/l18/index.htm
[Kowalski 1997] Gerald Kowalski: Information Retrieval Systems - Theory
and Implementation. Kluwer Academic Publishers, Boston, 1997
[Woodruff 1998] Allison Woodruff: CS260 Discussion of Information
Visualization. UC Berkeley, 1998.
© 2001 Franz J. Kurfess
Knowledge Presentation 86
Important Concepts and Terms













agent
automated reasoning
belief network
cognitive science
computer science
hidden Markov model
intelligence
knowledge representation
linguistics
Lisp
logic
machine learning
microworlds
© 2001 Franz J. Kurfess







natural language processing
neural network
predicate logic
propositional logic
rational agent
rationality
Turing test
Knowledge Presentation 87
Summary Chapter-Topic
© 2001 Franz J. Kurfess
Knowledge Presentation 88
© 2001 Franz J. Kurfess
Knowledge Presentation 89
Descargar

CSC 480: Artificial Intelligence