Improving the Quality of Visual Web Browsing by Using Weighted Graph Drawing - PowerPoint PPT Presentation

1 / 29
About This Presentation
Title:

Improving the Quality of Visual Web Browsing by Using Weighted Graph Drawing

Description:

http://sally.it.uts.edu.au/~maolin/gallery/tsakalo/map.html. 9/19/09. 10 ... Translation into Pictures. Geometrical mapping. Visual mapping ... – PowerPoint PPT presentation

Number of Views:54
Avg rating:3.0/5.0
Slides: 30
Provided by: mao8
Category:

less

Transcript and Presenter's Notes

Title: Improving the Quality of Visual Web Browsing by Using Weighted Graph Drawing


1
Improving the Quality of Visual Web Browsing by
Using Weighted Graph Drawing
Mao Lin Huang Department of Computer
Systems, UTS, NSW 2007, Australia.
2
Abstract
Web Site Maps (static dynamic) provide an
on-line assistance for users to browse through
the local web-spaces. Visual Web Browser allows
the user to graphically snoop the World Wide Web
via a sequence of successive visual maps of web
structure. However, these techniques only focused
on the problem of representing hyper-documents
and hyper-links. They did not consider the
problem of visually representing the attributes
that are associated with the hyper-documents
and hyper-links. This research tries to use
Weighted Graph Drawing methods to address the
problem of visually representing the attributes
of web pages, including the Connectivity, the
Access Frequency and the Depth of a page in the
web locality. Thus, the user can gain the
knowledge of importance of nodes while he/she is
browsing through the web-space.
3
General problems of web structure
  • Web site maps provide an overview diagram of the
    contents of a web site. There are four main
    concerns that site maps try to address.
  • Disorientation
  • Absence of physical context
  • Lack of control over distributed sites
  • Cognitive overhead

4
Web site mapping techniques
Since web site mapping is essentially
visualization process of information content of a
web site, various approaches are adopted based on
human visualization and perception capabilities.
Each approach or technique for web site mapping
has adopted one or a combination of these
capabilities hoping to exploit them to help the
user in navigation and comprehending the contents
5
Web site mapping techniques
  • From the creation point of view, the web site
    mapping techniques could be categorized into two
    main approaches
  • static mapping and
  • dynamic mapping.

6
Static mapping
Static mapping Static maps are designed by
studying the contents of the web site manually
and then determining the grouping of web pages
and giving them some pictorial or visual
representation on the web site map. Once the map
is prepared and installed it does not change
until the web site administrator updates or
changes it.
7
Web site mapping techniques
1D hierarchy
8
Web site mapping techniques
2D hierarchy
9
Dynamic Mapping
Dynamic maps are produced "on the fly. When
users navigate the contents of a web site(s), the
system automatically creates a sequence of maps
corresponding to the contents of the web site.
This makes it possible to always get the latest
(most recent) version of the web structure for
visualization. Dynamic mapping employs
mathematical optimization techniques, artificial
intelligence and heuristic algorithms to
automatically determine the best possible
visualization for the site map. A dynamic web
site appears differently from time to time
depending on the current contents of the web
site, and the interest of a particular user.
Example http//sally.it.uts.edu.au/maolin/gall
ery/tsakalo/map.html
10
Dynamic site mapping techniques
  • From visualization point of view, there are two
    categories of visualization techniques that can
    be used for dynamic web site mapping. They are
  • Focus context
  • Zooming filtering

11
Web site mapping techniques
Focus context (local navigation) The focus
context concept has some applications, such as
fisheye view Sarkar Brown, bifocal view
Pilgrim et al. 96 and hyperbolic tree Lamping
et al. 1995. It allows a user to view a small
central focus in detail while maintaining the
visibility of a large overall context or part of
the overall context. WebCutter Maarek et al.
97, WebMapper 11 are good examples of adopting
"focus context" concept for web site mapping.
12
Web site mapping techniques
Zooming filtering (local global) The zooming
filtering takes the form of selecting a subset
of the data along a range of numerical values of
one or more dimensions. It works by reducing the
amount of context in the display this
distinguishes them from the focuscontext
techniques, which attempt to retain all the
contextual information even if it must drawn so
small as to make it virtually invisible.
WebOFDAV, ThinkMap, HotSauce and WebMapper are
examples of adopting "zooming filtering"
concept for web site mapping.
13
Web site mapping tools
MAPA's 3D map was developed by Dynamic Diagrams
which implements site maps as Java applets.
MAPA adopts the 2D/3D hierarchical mapping"
method to show the hierarchical view of a web
site. It also uses "focus context" concept to
re-centering the site based on user's current
focus.
14
Web site mapping tools
WebMapper is developed by Netcarta. It creates a
hyperbolic view Lamping et al. 1995 of the
web site outline. This graphic is a tree diagram
modified by "fish-eye" view distortion
Figure 4 WebMapper uses hyperbolic tree viewing
that adopts both "focus context" "zooming
filtering" concepts for visualization. http//www.
inxight.com/products_wb/ht_server/ht_server_demos.
html .
15
Web site mapping tools
ThinkMap is developed by Plumb Design. It has the
application of Visual Thesaurus that accesses
data from the WordNet database, a publicly
available resource developed by the Cognitive
Science Laboratory at Princeton University.
3D ThinkMap adopts a 3D/2D force-directed layout
algorithm "zooming filtering" concept for
visualizat. http//www.visualthesaurus.com/
16
Web site mapping tools
This was part of my PhD work
WebOFDAV adopts a force-directed layout algorithm
"zooming filtering" concept for
visualization. It produces maps "on the flycan
be used for global web navigation. http//www.cs.n
ewcastle.edu.au/mhuang/spring_fifo/spring.html
17
Problem ?
However, these techniques described above did not
show the importance of each node (web page) in
the web locality. These visualization techniques
did not show the attributes of nodes and edges,
which are sometime very important to the user
where surfing through the web.
18
Attributes associated with nodes edges
  • Node (web page)
  • Connectivity (indegree outdegree)
  • Access frequency
  • depth (position in the file system)
  • medium type (html, gif, mic,nls, msc)
  • Edge (web link)
  • HTTP link
  • FTP link
  • MAILTO link


19
Metrics for measuring the importance of a node v
in the web locality
  • Connectivity
  • The importance of a node could be reflected by
    the connectivity of the node. If a node is
    connected to more other nodes in the web space,
    then we say this node is to be more important in
    the web locality. To find out the connectivity of
    a node the hyperlink structure (the topology)
    among the web pages needs to be extracted.
  • We use two terms, outdegree and indegree to
    measure the connectivity of a
  • node. The outdegree of a node is the number of
    other nodes that can be directly
  • reached from this node with the graph theoretic
    distance of 1. Similarly, the
  • indegree of a node is the number of other nodes
    that can directly reach this node
  • with the graph theoretic distance of 1.

20
Metrics for measuring the importance of a node v
in the web locality
  • Access frequency
  • The importance of a node could also be reflected
    by the access frequency of the node. This means
    that if a node (page) had more visitors in a
    periodical time, then we say this node is to be
    more important in the web locality. This
    information can be obtained by analysis of the
    log files. For each node we may calculate the
    number of times it was accessed in the preceding
    month. To keep the access frequency uptodate this
    process needs to be done periodically.

21
Metrics for measuring the importance of a node v
in the web locality
  • Depth
  • The importance of a node could also be reflected
    by the position of the node in the hierarchical
    file system of a web site. Generally, when a web
    site is developed, the file system hierarchy is
    formed so that we assume that the important nodes
    are normally higher up in the hierarchy. For
    example, in the Faculty of Information Technology
    site at UTS, the top directory of the web site
    has the home page of the Faculty.

22
A weighted graph model

A weight W(?) is a non-negative real value
associated with a node ? of a graph. A
weighted graph G (V, E, W, W) consists of a
graph G (V, E), a weight W(?) for each ? ? V,
and a weight W(e) for each e ? E. Each W(?)
has a set of sub-weights, W(?) w?1, w?2, w?3
, and similarly for each W(e), we have W(e)
we1, we2, we3 .
23
Weighted graph drawing

The traditional Graph Drawing concerns the
problem of constructing geometric representations
of graphs. Most GD algorithms only concern the
geometrical attributes of nodes and edges. They
calculate the pixel coordinates (x?, y?) for each
node ?, and a pair of coordinates (x?, y?) (x
µ, yµ) for each edge e (µ, ?). Obviously, the
pure geometric drawing D is no longer appropriate
for representing this kind of weighted graphs.
We need to employ more graphic properties to
visually represent this kind of weighted graphs.
24
Weighted graph drawing

A weighted drawing of a graph G (V, E, W, W)
is not only a geometrical function D V ? R2
that associates a drawing D(?) to each node ? ?
V, but also a visual function P V ? Rm that
associates a pictorial representation P(?) to
each node ? ? V. Here m is the number of the
graphic properties that could be applied on the
visual object (rectangle) of ?. Similarly, for
each edge e ? E, we may have a geometrical
drawing D(e) and a pictorial representation P(e).
25
Graphic Properties

Nodes - size of the graphic entity -
background color - shape of the nodes -
brightness - highlight/shadows - Z-coordinate
at overlaps - . Edges - length -
thickness - brightness - Z-coordinate at
crossings - ...
26
Translation into Pictures
hypertext

relational structure representation
page
A1
A2
A3
Geometrical mapping
x
y
geometrical structure representation
node
Visual mapping
x
y
node
v1
v2
v3
target pictorial representation
The visualization model
27
Actual picture creation
Weights

Graphic properties
node-size node-color node-shape node-brightness no
de-highlight node-z-coordinate edge-length edge-th
ickness edge-brightness edge-z-coordinate ...
node connectivity node access frequency node
depth node medium type http link ftp
link mailto link
w1
w2
many to many relations
28
Calculating graphic properties for nodes
http//www-staff.it.uts.edu.au/maolin/gd2001/test
.html
29
Conclusion and future directions
Write a Comment
User Comments (0)
About PowerShow.com