Visual Web Mining by sir17308

VIEWS: 16 PAGES: 30

									       Visual Web Mining



   Done by:
Yakovlev Michael

                           16 januar 2003
1. General information
on vizualizations.

2.Visualization of Web
usage and structures.

3.Visual Web.
1. General information
on vizualizations.

2.Visualization of Web
usage and structures.

3.Visual Web.
               DEFINITION


                  Visual Data Mining
the process of discovering implicit but useful
 knowledge from large data sets using visualization
 techniques
                                          (E.R. Tufte, 1997)
                      DEFINITION
                         Vizualization:
The use of computer-supported, interactive, visual
 representation of data to amplify cognition. (Card, S.K.,
  J.D., Mackinlay & B. Shneiderman 1999)

Visualization can be dicribed as the mapping of data
 to visual form that supports human interaction in a
 workspace for visual sence making. (Card, S.K., J.D.,
  Mackinlay & B. Shneiderman 1999)

                    Information visualization:
The use of computer-supported, interactive, visual
 representation of abstact data to amplify cognition.
  (Card, S.K., J.D., Mackinlay & B. Shneiderman 1999)
       VISUAL PERCEPTION
Human perceptual abilities are remarkable at
 spotting orderly and unusual patterns.

Visual patterns are more likely to be remembered
 - Easy comparisons of information.
                    HISTORY
The first work in data graphics was made by
 Playfair (1786).
     He was the first to use abstract visual
 properties (like line and area) to represent the
 data visually (Tufte, 1983).
From 1990 every year IEEE Visualization
 Conferece (Institute of Electrical and Electronics Engineers).
    Non-profit, technical professional association
 of more than 377,000 individual members in 150
 countries.
        THE BENEFITS OF
        VISUALIZATIONS
Powerful tool in exploring or explaining
 information.
Can reduce time of analyzing the information.
Can show some new details, result.
Reduce the quantity of data.
Could be easily presented to non-experts.
Allow to work on a great amount of data.
Visual data representation: great
opportunity vs. great danger:
One of the diagrams of O-ring damage used to make a decision to launch Challenger

                        (Nielson, Hagen and Muller, 1997)
The main variables are not presented correctly:

                           temperature is shown
                            textually rather than
                            graphically
                           degree of damage is not
                            mapped onto graphical
                            scale
                           no legend
                           the conclusion is not
                            understandable
Scattergraph of O-ring damage index as a function of temperature (Tufte, 1997)
Correct presentation:

                        depict the relationship
                         between main variables
                         (damage-temperature)
                        different type of damage
                         are put into a single index
                        proposed temperature is
                         also put in
                        The graph clearly shows
                         that there is always
                         damage below 65*
                          (Tufte, 1997)
     www.
statsoftinc.com
1. General information on
vizualizations.

2.Visualization of Web
usage and structures.

3.Visual Web.
        WebViz: a tool for WWW
         Access Log Analysis
The designer is provided with a graphical view
 of a local data base and access patterns.
  (it means that designer can see not only the documents with
  users access logs, as in most analyzing programs, but also the
  hyperlinks traveled by user).
The designer becomes a graphical information
 about accesses and pathes taken by user through
 the database.
(Pitkow, 1996)
WebViz:
          links are representing the
           hyper-links between
           documents
          nodes represent separate
           documents
          different colors and
           thickness (of links and
           nodes) can show some
           different parametrs, for
           example recency and
           frequency of accesses
   DISK TREE VS. DOME TREE

Both visualization techniques are mapping large
 Web sites.

Provide a better understanding of current design
 and of set of users and could be helpful in
 building the alternative site design.
(Chi, P.Pirolli, J.Pitkow, 2000)
                                     at center- root node
                                     next levels are mapped to
DISK TREE:                            new rings expanding from
links (yellow) have many crossings    the center
                                     the amount of space given
                                      to each sub-tree is
                                      proportional to the
                                      number of leaf nodes it
                                      contains
                                               Limitations:
                                     data is lost: overlaying user
                                      paths occludes the
                                      underlaying structure of
                                      Web site
                                     useses 2D
               Links (yellow) are laid
DOME TREE:     along significant paths
               (orange), elemenating
               crossings



             Only 3/4 of disk is used
              and at each next level the
              disk is extruded along Z
              dimension
             The structure is expanded
              from 2D to 3D
                                              coarsened stratograms are
Coarsened stratograms                          merging trees of multiple
- a tool for analyzing                         paths into directed acyclic
                                               graph, embedding these in a
users access logs:                             state space
                                              the visual variables area and
                                               color shows support and
                                               shape shows different
                                               actions(transitions-exits)
                                              they can display the paths of
                                               many as well as of
                                               individual paths
                                              they provide navigation at
                                               different level of details
                                                  (zooming-unzooming)
B.Berendt``Detail and context in web usage mining:coarsening and visualizing sequences``
1. General information on
vizualizations.

2.Visualization of Web
usage and structures.

3.Visual Web.
       The implementation of
     visualizations into WWW:

To improve usability of WWW
70% of all searches are failures (Funke, 2002)
User interface fails, when he becomes
 hundreds or thousands hits(results)
As an experiment visual Web searching
 service were developed
Miner3d.com
Results:

           Results of a search are
           visualized as graphic
           objects positioned by its
           relevancy, colorized
           according to a documents
           domain and textured by
           most important text
           information (title,
           domain, text, size)
            Conclusion:

In my work I have presented some basic
information on vizualizations and also
poweful techuniques and tools for
visualization and analysis of the Web
structures and users acces logs, and showed
some posible upgrade of searching, when
the visualizations are applyed.
DISCUSSION

								
To top