1241 |
- images not good, can't generate infinite supply easily, too much |
- images not good, can't generate infinite supply easily, too much |
1242 |
semantics |
semantics |
1243 |
|
|
1244 |
- thus, use solid colors even though know not really good. |
- thus, use solid colors even though know solid colors not really recognizable. |
1245 |
|
|
1246 |
|
|
1247 |
|
Surely it would be hard to remember, e.g., 100 textures. |
1248 |
|
However, the user does not have to remember all the textures; |
1249 |
|
it suffices to learn the textures of the most often used documents |
1250 |
|
and distinguish them from the rest: |
1251 |
|
see Fig.~\ref{fig-zipf}. |
1252 |
|
Many studies of web cache statistics |
1253 |
|
have shown that for instance file popularity approximately follows Zipf's law |
1254 |
|
(see, e.g. [xxxref]) so that a small number of documents accounts |
1255 |
|
for most of the use. |
1256 |
|
|
1257 |
|
\begin{figure} |
1258 |
|
\centering |
1259 |
|
%\fbox{\vbox{\vskip 3in}} |
1260 |
|
\ifpics |
1261 |
|
\includegraphics[width=6cm]{zipfarea.1} |
1262 |
|
\fi |
1263 |
|
\caption{ |
1264 |
|
\label{fig-zipf} |
1265 |
|
Zipf's law concretized: why remembering 15 textures helps. |
1266 |
|
In real life, accesses to documents often follow Zipf's law, meaning that |
1267 |
|
some documents get accessed far more often than most. |
1268 |
|
Each square represents a document, and the area of each square is scaled |
1269 |
|
to its rate of accesses. |
1270 |
|
The diagram shows 2000 documents weighted with Zipf's law with exponent 1.1. |
1271 |
|
The 15 most important documents account XXX\% of the accesses. |
1272 |
|
} |
1273 |
|
\end{figure} |
1274 |
|
|
1275 |
|
Thus, the experiment measured the recognition of only 15 textures, |
1276 |
|
and |
1277 |
|
|
1278 |
|
Furthermore, because the texture appearance has no correlation |
1279 |
|
with the document content, |
1280 |
|
the textures of any two important files are similar only by chance. |
1281 |
|
|
1282 |
We have conducted an experiment in a similar setting comparing |
We have conducted an experiment in a similar setting comparing |
1283 |
the recognition performance of our textures with solid color backgrounds. |
the recognition performance of our textures with solid color backgrounds. |
1318 |
\end{center} |
\end{center} |
1319 |
\caption{ |
\caption{ |
1320 |
\label{tableresults} |
\label{tableresults} |
1321 |
(percentage of trials and reaction time in seconds): |
Results of a recognition experiment with 15 previously seen |
1322 |
|
textures light colors to be picked out from 15 not previously seen |
1323 |
|
instances. |
1324 |
|
The numbers give the |
1325 |
|
percentage of trials and the average reaction time in seconds for |
1326 |
|
the five subjects. |
1327 |
} |
} |
1328 |
\end{table} |
\end{table} |
1329 |
|
|
1338 |
The observed difference in recognition performance is statistically |
The observed difference in recognition performance is statistically |
1339 |
significant [F(1,4) = 19.0, p = .012]. |
significant [F(1,4) = 19.0, p = .012]. |
1340 |
|
|
|
Of course, the experiment measured the recognition of only 15 textures, |
|
|
and surely it would be harder to remember, e.g., 100 textures. |
|
|
However, the user does not have to remember all the textures; |
|
|
it suffices to learn the textures of the most often used documents, |
|
|
see Fig.~\ref{fig-zipf}. |
|
|
Many studies of web cache statistics |
|
|
have shown that file popularity approximately follows Zipf's law |
|
|
(see, e.g. [xxxref]) so that a small number of documents accounts |
|
|
for most of the use. |
|
|
|
|
|
\begin{figure} |
|
|
\centering |
|
|
%\fbox{\vbox{\vskip 3in}} |
|
|
\ifpics |
|
|
\includegraphics[width=6cm]{zipfarea.1} |
|
|
\fi |
|
|
\caption{ |
|
|
\label{fig-zipf} |
|
|
Zipf's law concretized: why remembering 15 textures helps. |
|
|
In real life, accesses to documents often follow Zipf's law, meaning that |
|
|
some documents get accessed far more often than most. |
|
|
Each square represents a document, and the area of each square is scaled |
|
|
to its rate of accesses. |
|
|
The diagram shows 2000 documents weighted with Zipf's law with exponent 1.1. |
|
|
The 15 most important documents account XXX\% of the accesses. |
|
|
} |
|
|
\end{figure} |
|
|
|
|
|
Furthermore, because the texture appearance has no correlation |
|
|
with the document content, |
|
|
the textures of any two important files are similar only by chance. |
|
|
|
|
1341 |
%Our experience shows that at least the most recurring textures |
%Our experience shows that at least the most recurring textures |
1342 |
%are not only recognized but can |
%are not only recognized but can |
1343 |
%also easily be associated with the document content. |
%also easily be associated with the document content. |