Seek and You Shall Find

1

• Seek and You Shall Find
• Visualization Tools
• Professor Elaine Ferneley

2
What is Visualization?

• Defn (vb) to form a mental image or vision of
  .. to imagine or remember as if actually seeing
  
• Graphic representations of data can be used to
  make the information contained more accessible.
• The graphical form is easier to investigate than
  tables of data.
• It helps the user to search for items of interest
  and to detect patterns. Examples include graphs
  and bar charts.
• Cognitive activity it goes on in the mind,
  creation of an internal model or cognitive map
• Internal to the individual human being provides
  insight and understanding
• Individuals will interpret visualizations in
  different ways

3
Visualization Metaphors

• All attempts to create a representational picture
  from some information are visualization.
• Sometimes the representation will explain a
  physical fact e.g. a bar chart for the heights of
  all the children in a class.
• Sometimes the representation will explain
  conceptual facts e.g. business processes
• Information visualization is different because
  the mapping is not based on the physicality of
  the data.
• All data has a physical form words in a book,
  documents stored on a hard disk but in order to
  graphically display the meaning behind the data a
  new physicality has to be created.
• This new physical model creates a metaphor to
  help the user navigate the information that is
  presented to them. In Microsoft Windows, a
  desktop metaphor supports access to computer
  files.

4
A Classic Visualization
Individuals all have a unique understanding
perspective
5
Minards Map of Napoleons March on Moscow
Originally created in 1812

• Depicts several variables in a small space
• the size of the French army depicted by the width
  of the bands
• its location on a two-dimensional surface
• the direction of the movement of the advance
  (pink upper band) and retreat (black lower band)
  
• the temperature on certain dates during the
  retreat.

6
Florence Nightingales Coxcombs to highlight
needless deaths during the Crimea War (1854-56)

• Mortality peaked in January 1855
• - 2761 died of contagious diseases,
• 83 of wounds, 324 of other causes.  
• Based on the army's average
• strength of 32393, Nightingale
• computed an annual mortality rate
• of 1174 per 1000. 

• Blue deaths from preventable
• disease (cholera typhoid)
• Red deaths from wounds
• Grey deaths from other causes

7
Key Visualisation Issues

• Selection
• should all data be represented
• can selection take place automatically
• Is it useful to suppress information.
• Presentation
• How do you lay the data out e.g. Harry Becks
  distortion of the tube network to make the best
  use of space and make it more memorable
• New issues of how to present on mobile devices.

• Representation
• Colours, lines, slopes, graphs, pie charts etc
• Can attributes be combined
• How do you represent many attributes (gt20 ?).
• Scale
• How to cope with vast quantities of data
  abstraction.
• Rearrangement, interaction and exploration
• Can the user view be changed
• Is it possible to explore the underlying
  datasets.

8
Rearrangement the key to insight

• Simple example for illustration
• 10 crops e.g. rice, cereal, barley, corn etc.
• 7 treatments e.g. fertiliser, insecticide,
  pruning etc.
• Result of treatment noted
• Purple improvement
• White degradation.

7 treatments fertiliser, insecticide, pruning
10 crops, rice, barley, corn
9
Rearrangement the key to insight
7 treatments fertiliser, insecticide, pruning
A B C D E F G
1 3 8 2 6 10 4 7 9 5
10 crops, rice, barley, corn
10
Rearrangement the key to insight
7 treatments fertiliser, insecticide, pruning
A B C D E F G
A D C E G B F
10 crops, rice, barley, corn
Can now clearly see that certain groups of
treatments are appropriate for certain types of
crops
11
Left and Right Brain Activity

• The need to produce visualisations of data can be
  seen as the need to incorporate left and right
  brain activity.
• Current search engines are seen as analytical,
  methodical, serial technologies suited to the
  activities of the left hemisphere of the human
  brain.
• Shapes and patterns have been ignored by the text
  based approach and would tap in to right
  hemisphere activity.
• Bad displays of data obviously make perception
  much more difficult for people.
• The layout of the voting cards in some US states
  was seen as the main reason why the whole 2000
  election was thrown into turmoil because the poor
  representation of information actually hampered
  peoples decision making capabilities.

12
US Voting 2000 Why we have Bush??
13
US Voting Why we have Bush??
14
US Voting

• Errors associated with the design of the Palm
  Beach County ballot were primarily due to poor
  ballot layout, resulting in problematic spatial
  mappings. 
• The two-page format of the ballot violated the
  expectations of voters.  People reading English
  text read from left to right and will read a
  left-hand page from top to bottom before reading
  a right-hand page. 
• Thus, the natural behaviour for voters was to
  start at the top of the left-hand page and read
  down. 
• However, holes on the ballot book corresponded in
  alternating fashion to candidates on the left and
  right pages. 
• Some voters claimed to be confused and said that
  they wanted to vote for the second candidate from
  the top left (Gore) but punched the second hole,
  which actually corresponded to Buchanan, who was
  listed on the right-hand page

15
Is this Possible???
16
Focus on the dot in the centre and move your head
backwards and forwards
17
Are the purple lines straight or bent??
18
Do you see grey areas in between the squares?
Where did they come from?
19
You should see a mans face and a word
20
This is not animated!
21

Who do you see??

1. Relax and concentrate on the 4 small dots in the
   middle of the picture for 40 secs
2. Then, take a look at a wall near you (any smooth,
   single coloured surface)
3. You will see a circle of light developing
4. Start blinking your eyes a couple of times and
   you will see a figure.

22
Rudolph Arnheim the Power of the Center
23
When things come out of a centre or, the reverse,
bear in on a centre, a dynamic is created. Our
eye is drawn to or away from the centre
24
When things come out of a centre or, the reverse,
bear in on a centre, a dynamic is created. Our
eye is drawn to or away from the centre
25
DaVincis The Last Supper
26
What do You Feel?
27
What do You Feel?
28
What do You Feel?
29
What do You Feel?
30
Visualization and Large Datasets

• It is hoped that the previous experience of the
  user will help them with the new, abstract
  information.
• Investigating the semantics of data means
  creating links and classifications regardless of
  physical boundaries and it is these semantic
  connections that can be given a graphical
  representation.
• The aim of information visualisation is to
  provide a computer-based, interactive graphic of
  abstract data which improves cognition
• As an alternative to a search on the internet
  returning a text list, a graphical representation
  can show the links between sets of concepts.
• The aims are to allow the user to find what they
  need faster and to identify previously
  unrecognized relationships.

31
Problem of Information Retrieval from
intranets/the internet

• External factors
• Limited time, exponential growth of content
• increase in naïve users, few new tools
• Query factors
• Encourage fast wandering, bibbling, collaboration
• Support for fuzzy query formulation do you know
  what you want to know ???

32
Mapping Cyberspace using Geographic Metaphors

• IP access across USA
• Different colours represent time routing
• Not accessible

• IP address access from Stamford University

33
Statistical Maps of Cyberspace - Size of country
represents internet usage

• A "census" of Internet connectivity by countries
  has been developed at Computer Science
  Department, University of Wisconsin - Madison,
  USA.
• The map opposite shows the differential levels of
  network connectivity

34
The Art of Asking

• Volvo intranet study
• 58 single keyword searches, 34 single phrase
  searches, 8 gt one word or phrase
• Why are sophisticated questions not asked
• Users do not have mathematical or logic training
  Boolean expressions inappropriate
• ((nivana OR grunge) AND seattle) AND NOT
  hinduism
• Semantic gulf only 15 of users use the same
  expression to search for a specific topic
• Were lazy ! usually only submit one word
• My index is bigger than yours ! approach of
  most search engine suppliers

35
Internet usage in Europe

• Produced by Eric Guichard, at the Ecole Normale
  Superieure, Paris.
• Countries are colour-coded according to hosts per
  capita and the green circles show domains per
  capita.
• Blue diamonds show the national population.

36
Tile Based Maps

• "ET-Map" - a multi-level category map of the
  information space of over 100,000 entertainment
  related Web pages listed by Yahoo!.
• Developed by by Hsinchun Chen, at the University
  of Arizona, USA

37
Virtual World Representations

• The 3D cityscape view of the Web generated by
  Map.Net.
• You fly-through the world, with individual
  websites represented by different buildings.
• The large skyscrapers are the most popular and
  important site on the Web.

38
Themescape very advanced but can not be
reconfigured by interrogation

• a visual landscape of hundreds or thousands of
  web pages
• Peaks - represent concentrations of documents
  about a similar topic.
• more documents create higher peaks
• valleys between peaks contain fewer documents but
  with more unique content
• Documents in common mountain ranges or valleys
  are related.
• Topic Labels - reflect the major two or three
  topics represented in a given area of the map,
• quick indication of what the documents are about
• additional labels often appear when you zoom into
  the map for greater detail.

39
Hyperbolic Trees www.webbrain.com

• Ideal for hierarchical data
• Search engine results
• File directories
• Data, information and certainly knowledge is not
  hierarchical
• Extension of the hyperbolic tree should consider
  cross branch associations.

40
Example 2 Spectacle from aidministrator.com

• Data represented as a system of text labels
  spheres connected by straight lines 
• Each label represents a class (category) of
  pages, e.g. sports 
• Each sphere is an instance of a class (every page
  belonging to that category). 
• The lines indicate that an instance is a member
  of a class or that a class is a subclass of
  another one (e.g. football is a member of sports
  and sports could be a subclass of entertainment).

41
How to Read a Spectacle Graphic

• All elements are located in space through a
  system of attraction and repulsion between the
  objects (as if you had springs inside the lines)
• objects that are semantically close appear
  spatially close
• objects that are semantically far away appear in
  distant locations
• In this context "semantically close" means that
  two classes share many instances or that two
  instances belong to the same class.

42
How to Read a Spectacle Graphic Cont.

• Example depicts the contents of recruitment
  agency database
• Jobs classified economic sector recreation,
  finance, education
• The size intersections of the classes stands
  out immediately due to the clustering of spheres
• quite a few of the jobs are classified under
  several classes e.g. one of them "hangs" from
  finance, management secretarial.
• Classes placed on opposite sides of the diagram
  don't have any member in common (government
  security vs., healthcare sports).

43
Support Mechanisms

• Natural language interfaces (AskJeeves
  www.ask.com)
• show me info on Nivana or other grunge bands
  from Seattle but nothing on Hinduism
• Query expansion
• Query is automatically augmented with synonym
  keywords from a thesaurus
• Tends to increase the amount of data brought back
• Domain pruning
• Predefined categories e.g. Yahoo!
• Intelligent agents
• Learn about the user
• Support user collaboration

44
Exploration - Understand/identify the reasons
behind a rearrangement

• Must be interactive
• Reconfigurable interfaces
• Multiple views/perspectives
• Must be capable of interrogation
• Embedded query manipulation languages
• Current limitations on recreation of graphical
  interfaces quickly
• Key questions
• Is the interface intuitive
• Can the interface be reconfigured to give a user
  specific view
• Can multiple concurrent views be displayed
  essential for knowledge sharing
• Can the underlying data be queried
• Can alternative views be generated within an
  acceptable time period.

45
Finally, not All Visualizations have to be
Serious!
46
Summary
Visualisation is not a new discipline