The latest issue of Technology Research News (TRN) contains a short item about how photos can make icons meaningful. Researchers from Northwestern University are working on semanticons which will help us to quickly identify files. After analyzing a file name and its contents, their system uses a database of pictures to generate a semanticon, a specific icon based on the file type or the folder where it resides. Apparently, early users were able to navigate through their directories 20% faster than when looking at regular icons.
Semanticons can enhance the representation of files by offering symbols that are both meaningful and easily distinguishable. The semantics of a file is estimated by parsing its name, location, and content to generate a context, which is used to query an image database. The resulting images are simplified by segmenting them, computing an importance value for each segmented region, and removing unimportant regions. The abstract look-and-feel of icons is achieved using non-photorealistic techniques for image generalization.
Before going further, let's look at several examples.
This first example shows semanticons generated for different file types (respectively comments.doc, labTests.ppt, Estimate.xls, silhouette_algorithms.ppt and party.html) (Credit: Northwestern University).
And in this second one, here are the semanticons generated for different files carrying the same name (article.rtf) but stored in various locations. In this case, the parent directory influences the semanticon creation (Credit: Northwestern University).
Finally, this third example illustrates how the composition process is influenced by the file name. In this case, you can see semanticons generated for two files named respectively "java_article.rtf" and "article_java.rtf" (Credit: Northwestern University).
But is this approach efficient? The researchers ran two preliminary studies which conclude that "semanticons decrease the time necessary to locate a file in a visual search task and enhance performance in a memory task."
[In the first study, there were] two phases, conducted such that each participant first completes the common icon visual search, followed by the semanticon visual search. [And] the semanticons were recognized on average 1.96 seconds faster than icons.
The second study was a memory game where players are in front of a grid of randomly sorted cards placed faced down. "The goal is to turn over two cards at a time. If a match is made, then the cards are removed. Otherwise, the cards are placed face down and another set of cards are turned over. The game ends when all of the pairs are found."
There were two sets of trials, one with common Macintosh OS X icons, the other with semanticons. And the results are clearly better with semanticons: 88 seconds to finish the game to compare with 112 seconds with standard icons.
For more information, you should read this technical paper, "Semanticons: Visual Metaphors as File Icons" (PDF format, 10 pages, 2.20 MB), from which the above images and some quotes about the studies have been extracted.
So will semanticons be adopted by the software industry anytime soon? Probably not, because there are no standards behind this initiative. But it could be helpful for an individual who wants to get his own set of individual icons.
Sources: Technology Research News, December 19, 2005; and various web sites
You'll find related stories by following the links below.