/[gzz]/manuscripts/Paper/exper.txt
ViewVC logotype

Diff of /manuscripts/Paper/exper.txt

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

revision 1.3 by jvk, Wed Feb 26 10:58:48 2003 UTC revision 1.4 by jvk, Wed Feb 26 15:06:43 2003 UTC
# Line 1  Line 1 
   
   
 Test cases:  
   
 - text size: small / big  
 - bg type cases: color / shapes, i.e.,  
      - white  
      - one color  
      - full texture  
      - grayscale texture (intensity scale "compressed" to a certain range)  
   
1  General:  General:
2    
3  - 6 subjects do all releavant cases for each test  - 6 subjects do all releavant cases for each test
4  - the order of the test cases is random for each subject  - the order of the test cases is random for each subject
5    to componsate memory effects    to componsate memory effects
6    
 Recognisability test (usefulness)  
 ---------------------------------  
   
 - objective: show that the background textures are useful  
 - means: show that fragments of documents can be recognized much  
   better when unique backgrounds are used  
   
      - buoy view testing XXX  
   
   
   
   
 Suppose that the first page of each of hundreds of documents is torn into  
 pieces and the pieces are put in a single container.  
 It is intuitively clear that the pieces can be identified and put  
 back together much more rapidly if each document has a unique  
 background texture as compared to, e.g., single background colors.  
   
 Another question is whether the extra features of the textures help  
 recognition as much if the fragments cannot be seen simultaneously  
 for comparing.  
   
 "Applicability"(better word?) test  
 ----------------------------------  
   
 - objective: show that the textures are better than single colors  
   
     - test cases: single color / full textures [ / grayscale textyres ]  
   
     - show 30 textures in series and then test recognisability with  
       30 new mixed in and shuffled with the 30 old, shown in series  
   
 XXX: small-scale fragments do lose shape features.  
 what to say about this  
   
7  Readability test (non-disruptiveness)  Readability test (non-disruptiveness)
8  -------------------------------------  -------------------------------------
9    
# Line 88  ISSUES: Line 42  ISSUES:
42       - is the total of 2 * 6 subjects = 12 trials for each       - is the total of 2 * 6 subjects = 12 trials for each
43         (font size, test case) pair enough to compensate the         (font size, test case) pair enough to compensate the
44         memory effect and to obtain small enough variance?         memory effect and to obtain small enough variance?
45    
46         - how to scale the font?
47           should the background be scaled with it?
48         - what textures to use?
49           If we use unique textures for each text, they add extra
50           variation to the reading speed.
51    
52    Recognizability test
53    --------------------
54    - objective: show that textures can be remembered and that
55      they are better than single colors
56    
57        - test cases: single color / full textures [ / grayscale textyres ]
58    
59        - learning phase: show 30 textures in series, 5 seconds each
60    
61        - recognition phase:
62          trials: 30 new textures mixed in and shuffled with the 30 old;
63          for each trial, the subject answers old/new
64    
65    ISSUES:
66        - probably need some repetition to be able to memorize!
67    
68    
69    
70    
71    
72    
73    SCRATCH
74    =======
75    
76    
77    
78    Test cases:
79    
80    - text size: small / big
81    - bg type cases: color / shapes, i.e.,
82         - white
83         - one color
84         - full texture
85         - grayscale texture (intensity scale "compressed" to a certain range)
86    
87    
88    XXX: small-scale fragments do lose shape features.
89    what to say about this
90    
91    Recognisability test (usefulness)
92    ---------------------------------
93    
94    - objective: show that the background textures are useful
95    - means: show that fragments of documents can be recognized much
96      better when unique backgrounds are used
97    
98         - buoy view testing XXX
99    
100    
101    
102    
103    Suppose that the first page of each of hundreds of documents is torn into
104    pieces and the pieces are put in a single container.
105    It is intuitively clear that the pieces can be identified and put
106    back together much more rapidly if each document has a unique
107    background texture as compared to, e.g., single background colors.
108    
109    Another question is whether the extra features of the textures help
110    recognition as much if the fragments cannot be seen simultaneously
111    for comparing.
112    

Legend:
Removed from v.1.3  
changed lines
  Added in v.1.4

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