Feature Channels

The Dumpster_by: Golan Levin, Kamal Nigam and Jonathan Feinberg.

http://artport.whitney.org/commissions/thedumpster/

The dumpster is a visually dynamic, interactive, curiosity machine which makes excellent use of the feature channels discussed in class. The byline calls it "a portrait of romantic breakups collected from blogs in 2005". In effect, it's a multi-dimensional data visualization which uses color, size, motion, and spacial layout to display similarities in terminated relationships that were posted on the web in 2005.

These "breakups" are deemed similar by language used, demographic and emotional content and are represented by color, size and proximity. The multi-sized and colored circles displayed on the screen bounce into action once the Dumpster is initiated. Each circle is representative of the overall data, some 20,000 entries. When you scroll over the circles, the cursored one, changes color from orangey-red to pink with a blue surround, as well as increasing in size. This is effective in differentiating between what can be selected and the rest of the bouncing, vibrating circles. Once selected the circle turns yellow with an orange surround and increases in size as well. At this point several colored pixels appear to the left of the screen. This shows similar breakups within the overall data, with the brighter pixels being closer in search criteria: language used in the breakup; sex and age of the author; and "emotive words" (?). Further as soon as a circle is selected the circles that enter the center screen become the data represented to the left. The size of these circles shows the length of the post and the color and proximity to the chosen post relates to their similarity in search criteria.

The bar graph below the central screen of bouncing balls is a timeline which relates to the data on the left. It shows a timeline broken up by weekday, month and date and displays what could be  selected bar by enlarging the scrolled over bar within the graph as well as changing it's color. Above the enlarged bar is the weekday, month and date and the number to the left of the date represents the number of entries for that day which is further represented by the height of the bar within the graph. When scrolling over the data (pixels) to the left, the bar graph data becomes active, displaying the details of this macro view, while the blue box under the cursor shows the general area of pixels which are being scrolled over. The ID number, age, sex, and number of matches of the active pixel are displayed in the bottom left corner along with a blow-up of the area.

Super interesting project. Too bad my interest in the content is really limited. The interactivity is great except that the data can't be chosen, for some reason (?), from the timeline. This seems like an obvious thing to do. Then I could see how many people were dumped on the same days I was!  The overall display is visually engaging to the point of turning my mind to mush just watching these balls do their thing, making effective use of read and discussed "feature channels".