/[gzz]/manuscripts/Paper/paper.tex
ViewVC logotype

Diff of /manuscripts/Paper/paper.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1.74 by jvk, Sun Mar 30 20:09:40 2003 UTC revision 1.75 by tjl, Mon Mar 31 04:36:52 2003 UTC
# Line 59  PREATTENTIVE FEATURES IN FEATURE VECTOR! Line 59  PREATTENTIVE FEATURES IN FEATURE VECTOR!
59    
60  TJL  TJL
61    
62  We apply a rough, qualitative model of visual perception  We present a perceptually designed hardware-accelerated
63  to motivate  algorithm for generating unique background textures for data.
 general principles for designing recognizably unique textures  
 for use as backgrounds for data.  
64  To be recongizable,  To be recongizable,
65  the texture should produce a random feature vector in the brain  the texture should produce a random feature vector in the brain
66  {\em after} visual feature extraction.  after visual feature extraction.
   
 Our motivating example is...  
   
 We show how object identity can be visualized  
 by  
 unique background textures  
 procedurally generated from the identity (e.g., hashcode) of the document or data item.  
   
 XXXWAFFLE Unique backgrounds can assist  
 user orientation  
 when browsing a set of objects with  
 similar overall appearance,  
 especially in focus+context views.  
67    
68    Our motivating example is the BuoyOING user interface for
69    browsing hyperlinked document sets with fluid, non-disruptive linking.
70    The interface shows a fragment of the target document of a link in the marginal,
71    which, upon traversing the link, expands to fill the screen.
72    Our goal is to avoid user disorientation by
73    texturing each document with a unique background so that the originating
74    document can easily be recognized from the fragment.
75    The user should then be able to learn the textures of the
76    most often visited
77    documents, as per Zipf's law.
78    
79    %We show how object identity can be visualized
80    %by
81    %unique background textures
82    %procedurally generated from the identity (e.g., hashcode) of the document or data item.
83    
84    %XXXWAFFLE Unique backgrounds can assist
85    %user orientation
86    %when browsing a set of objects with
87    %similar overall appearance,
88    %especially in focus+context views.
89    %
90  %Unique backgrounds could be useful for assisting user orientation  %Unique backgrounds could be useful for assisting user orientation
91  %in several different user interfaces; our primary application  %in several different user interfaces; our primary application
92  %is navigating hyperstructures using Focus+Context views.  %is navigating hyperstructures using Focus+Context views.
93  %  %
94  %We can rapidly generate a texture for any document the user visits,  %We can rapidly generate a texture for any document the user visits,
95  %and the user will be able to learn the textures of the  %and
96  %most often visited  
 %documents, as per Zipf's law.  
97  %  %
98  %In Focus+Context views, the textures can act as visual cues in the context  %In Focus+Context views, the textures can act as visual cues in the context
99  %(information foraging).  %(information foraging).
100    
101  We discuss our freely available hardware-accelerated implementation  We discuss our freely available hardware-accelerated implementation
102  of unique backgrounds  of unique backgrounds
103  on the NV10 and NV25 architectures, and  on the NV10 and NV25 architectures.
104  show an example user interface for browsing linked PDF documents  
105  in a focus+context view using unique backgrounds.  We show the results of an initial experiment ... XXX
106    
107  Text readability is a major concern for using such backgrounds,  Text readability is a major concern for using such backgrounds,
108  and we discuss a method for enhancing readability by unnoticeably  and we discuss a method for enhancing readability by both providing
109  bleaching the background around text..  fast, interactive zooming and
110    unnoticeably bleaching the background around text.
111    
112  % The implementation works by combining a small set of basis textures  % The implementation works by combining a small set of basis textures
113  % and perceptually chosen colors  % and perceptually chosen colors
# Line 177  have similar appearances. Line 184  have similar appearances.
184  % data identity  % data identity
185  % to make  % to make
186  % similar but distinct objects distinguishable and easily recognizable.  % similar but distinct objects distinguishable and easily recognizable.
   
 Discrete, unordered variables often occur coupled to other  
 variables when drawing graphs; for example, a variable representing  
 different makes of cars would be such.  
 If the number of different values that the variable takes are few,  
 distinct symbols can be used...  
   
 Visualizing a discrete variable with unordered values  
   
187  In this article, we introduce the use of procedurally generated unique backgrounds  In this article, we introduce the use of procedurally generated unique backgrounds
188  as a visualization of data identity: if each data item with a different identity has  as a visualization of document identity: if each document has
189  a different, easily distinguishable texture, the user can become aware of the identity  a different, easily distinguishable background texture, the user can become aware of the identity
190  of a displayed item at a glance, without explicitly reading the title.  of a displayed item at a glance, without explicitly reading the title.
191  Even more importantly, the user can become aware of the identity just by seeing  The user can even become aware of the identity just by seeing
192  any {\em fragment} of the item, instead of the ``title page''.  any {\em fragment} of the item, instead of the title page. This property is vital
193    for our example application discussed in Section~\ref{secbuoyoing}.
194    
195    
196  % as a navigation aid in focus+context views.  % as a navigation aid in focus+context views.
# Line 212  any {\em fragment} of the item, instead Line 211  any {\em fragment} of the item, instead
211  % a more prominent target for tracking movement between views.  % a more prominent target for tracking movement between views.
212    
213  In the following sections,  In the following sections,
214  we first review related work on texturing,  we first review related work on texturing.
 Focus+Context views.  
215  Next, we discuss the motivating example for this work:  Next, we discuss the motivating example for this work:
216  a specific (xupdf XXX) focus+context  the BuoyOING focus+context
217  user interface to a hypertext structure.  user interface to a hypertext structure.
218  Then, we formulate general principles for designing  Then, we formulate general principles for designing
219  recognizable backgrounds and present a hardware-accelerated implementation.  recognizable backgrounds and present a hardware-accelerated implementation.
220  Following this, we discuss enhancing text readability on such backgrounds  Following this, we discuss enhancing text readability on such backgrounds
221  and practical experiences.  and practical experiences.
222  Finally, we show an example application of unique backgrounds  %Finally, we show an example application of unique backgrounds
223  for browsing linked PDF documents in a focus+context view.  %for browsing linked PDF documents in a focus+context view.
224    
225  \section{Related work}  \section{Related work}
226    
 \subsection{Texturing}  
   
227  The {\em texture} of a surface, taken literally, is its translation-invariant statistical microstructure.  The {\em texture} of a surface, taken literally, is its translation-invariant statistical microstructure.
228  In computer graphics,  In computer graphics,
229  the word {\em texturing} is used in  the word {\em texturing} is used in
# Line 334  and statistically, as samples from a pro Line 330  and statistically, as samples from a pro
330  %dimensionality.  %dimensionality.
331  %% XXX: this is something we should experiment with our textures  %% XXX: this is something we should experiment with our textures
332    
 \subsection{Focus+Context views}  
   
 Focus+Context, or, fisheye views\cite{fc-fisheye} are  
 a paradigm for viewing large,  
 structured information sets  
 by showing the current area of  
 interest (focus) magnified  
 and the structurally connected but further-away  
 elements peripherally, with less magnification.  
 Much of the work on focus+context views has  
 concentrated on tree structures\cite{lamping96hyperbolic,fc-images},  
 or flat 2D images or maps\cite{fc-taxonomy}.  
333    
334  % The type of focus+context view for whi  % The type of focus+context view for whi
335    
# Line 439  or flat 2D images or maps\cite{fc-taxono Line 423  or flat 2D images or maps\cite{fc-taxono
423    
424  \section{The motivation for Unique Backgrounds: the BuoyOING user interface}  \section{The motivation for Unique Backgrounds: the BuoyOING user interface}
425    
426    \label{secbuoyoing}
427    
428    Focus+Context, or, fisheye views\cite{fc-fisheye} are
429    a paradigm for viewing large,
430    structured information sets
431    by showing the current area of
432    interest (focus) magnified
433    and the structurally connected but further-away
434    elements peripherally, with less magnification.
435    Much of the work on focus+context views has
436    concentrated on tree structures\cite{lamping96hyperbolic,fc-images},
437    or flat 2D images or maps\cite{fc-taxonomy}.
438    
439  TJL  TJL
440    
441  The motivating example for  The motivating example for
# Line 1292  for most of the use. Line 1289  for most of the use.
1289  \fi  \fi
1290  \caption{  \caption{
1291  \label{fig-zipf}  \label{fig-zipf}
1292  Zipf's law concretized: why remembering 15 textures helps.  Zipf's law concretized: why distinguishing 15
1293    textures from a large number of others helps.
1294  In real life, accesses to documents often follow Zipf's law, meaning that  In real life, accesses to documents often follow Zipf's law, meaning that
1295  some documents get accessed far more often than most.  some documents get accessed far more often than most.
1296  Each square represents a document, and the area of each square is scaled  Each square represents a document, and the area of each square is scaled
1297  to its rate of accesses.  to its rate of accesses.
1298  The diagram shows 2000 documents weighted with Zipf's law with exponent 1.1.  The diagram shows 2000 documents weighted with Zipf's law with exponent 1.1.
1299  The 15 most important documents account for 50\% of the accesses.  Here, the 15 most important documents account for approximately half of the accesses.
1300  % 0.50469672124463749  % 0.50469672124463749
1301  }  }
1302  \end{figure}  \end{figure}
# Line 1523  dimensionality. Line 1521  dimensionality.
1521  %  %
1522    
1523    
1524    Discrete, unordered variables often occur coupled to other
1525    variables when drawing graphs; for example, a variable representing
1526    different makes of cars would be such.
1527    If the number of different values that the variable takes are few,
1528    distinct symbols can be used...
1529    
1530    Visualizing a discrete variable with unordered values
1531    
1532    
1533  \section{Acknowledgments}  \section{Acknowledgments}
1534    
1535  The authors would like to thank John Canny,  The authors would like to thank John Canny,

Legend:
Removed from v.1.74  
changed lines
  Added in v.1.75

savannah-hackers-public@gnu.org
ViewVC Help
Powered by ViewVC 1.1.26