Synesketch is a Java library that translates text into visual representations of emotion. To implement this, I had to move from the Processing development environment to Eclipse. I've heard it's easy to transition, but apparently I've only heard LIES!
What the hell is "Ant" and why does it keep popping up when I try to run something? Why did it take 15 minutes to figure out how to turn on line numbers? What's a package and why is Eclipse hollering at me for using the default? What the hell is try and why do I keep having to use it to catch exceptions? Processing is fairly easy for designers with no programming background at all. Eclipse is probably easier to work in if you're a programmer but most designers I know will look at this program and its infinite buttons and unfamiliar words, go "wah" and move on.
That being said, I really, really, really wanted to play with this library so I managed to get it working, sort of. It broke when I tried to add text fields since I'm not sure how/where fonts need to be stored.
Above are the patterns for the six basic emotions (originally defined by Paul Ekman) included in the Synesketch package. I spent some time this morning inputting text to test the patterns to make sure they worked. Then I decided to "analyze" the home pages of a couple of news sites. I use analyze in quotes because I have no idea what kinds of algorithms are used and even if I did know I probably would not understand. That being said, look what came up with the below news sites!
Fox is angry as hell. Msnbc is a little bit scared, a little bit disgusted and a little bit saaaaad. It would be fantastic fun to create some kind of standalone or web app using this library. I'd like to run something that analyzed some set of websites every day and collected that data over time. Although now that I write that, it'd be possible to hook the library through the NYTimes api so we wouldn't have to wait to accumulate that data. Oh, internet, how much joy you bring.
Labels: data visualization, processing, programmingWednesday, September 2, 2009 // 0 Comments
One down, two to go. This is one of my final projects for my interactive studio. I figured I'd go ahead and document it before I forgot about it. Plus, I just got back a 42x30" poster and it's perty.
So this graph is the updated version of its little cousin. I just changed the proportions of the graph so it would fit in a traditional poster size (and was therefore cheaper to print), let the highest bars display off the chart so it was easier to see the smaller number of mentions, and rewrote the code a tad so it would display the number of mentions and the war mentioned on the actual bar as well as printing the months at the bottom of the print. I also removed the random opacity. It added a nice texture but it didn't really add any information. Instead, I made all the bars slightly transparent so it was possible to see other wars underneath it. Here's a close-up.
Yay! One project down. Two more to complete.
Labels: api, data visualization, interactive, print, processing, programming, school workWednesday, April 29, 2009 // 0 Comments
For our Interaction Scripting Studio final project, I am focusing on my data visualizations of war, films, media and probably death counts (or maybe just make puppy viewer work dynamically).
So tons of websites with large amounts of data are making APIs available to developers for application development and visualization (Flickr, Twitter, NYTimes, pretty much any next gen website, etc). There are also sites that provide visualization tools for you and encourage experimentation. The NYTimes Viz Lab is one of them. They provide a set number of datasets and a number of tools (like Wordle) and let people go crazy. Google also provides tons of tools including their Google Visualization API. IBM's Many Eyes takes it a step further and allows users to add their own data sets.
It's an exciting field and it makes my brain buzz. However, I found this article in SEED Magazine Getting Past the Pie Chart that talks about how, perhaps, the data viz explosion may not be making data any clearer. There is also a very real danger of making causal connections where none exist.
Still, data visualization done well can be a combination of design and science which becomes beautiful, meaningful and inspiring. I absolutely love it. Here is a screenshot of a project I'm working on. It's the further refinement of one of my original circle graphs. What I'd like to do is see if I can add some more stats to the graph without overloading it. In my head that means tilting the graph back to add a third dimension and perhaps including casualty data vertically from each war (and if I can find it, for each movie the films are about).
Labels: api, data visualization, graphs, interactive, processing, programming, school workSaturday, April 18, 2009 // 0 Comments
More data visualizations. These all use Processing. The data I eventually was able to scrape includes info from IMDB and the NYTimes API. The first one pulls total instances of the wars as keywords in NYTimes articles from 1981-2008 (they don't have anything indexed before 1981 at the moment), then it spits it out into a long bar chart. The large orange spike is the beginning of the Gulf War. The subsequent yellow/orange spikes are 9/11 and the brownish color spike is the start of the Iraq War. I have no idea what the strange WWII peak around 10/95 is.
This visualization uses both the NYTimes Articles API and the scraping from IMDB. It is a comparison between word counts in the NYTimes abstracts where WWII is a keyword and the descriptions on IMDB of WWII movies. I haven't actually looked at this that much as I just wrote it today (this stuff is all due tomorrow) but the first thing I notice is that in "reality" (the NYTimes articles) Japan features much larger than Germany whereas in the movie descriptions, Germany shows up more often.
These last two visualizations are of the same data set. The most popular movies about wars (according to IMDB) from 1900-2008.
Labels: data visualization, interactive, processing, programming, school workThursday, March 12, 2009 // 0 Comments
For my interactive studio, we're working on data visualizations. As we're only going over it for a couple of weeks, we're doing really basic stuff. I've decided to look at movie data and so far I've got a couple of draft visualizations.
This first one is a simple plot of the top 20 most popular movies (according to IMDB) from various genres over time. Though the colors are too dark, the graph shows basic trends like the popularity of Westerns in the 60s or how more recent animation is much more popular than, say, Fantasia.
I also took the roles of Oscar winning actresses and actors and grouped them into categories so we could see media-based gender roles. This is a much looser interpretation of the data (my categories are obviously somewhat arbitrary based on what I saw as trends) but it's an interesting study nonetheless. As these sketches are done in Processing they take a bit of time for me and my non-programmer brain so I've only fully completed the actresses sketch but here's a screenshot.
This assignment is quite a bit of fun and while the math and programming for most of the beautiful visualizations at places like Visual Complexity is beyond me at the moment, I hope to do some hand drawings, some more intricate illustrations in Illustrator and perhaps a physical data representation of some sort.
Labels: data visualization, interactive, school workThursday, February 26, 2009 // 0 Comments
BACK TO THE TOP
The quote at the top of this page is from the March 25, 1893 Newark Daily Advocate via Nick de la Mare..
• Data Visualization
• School Work
• January 2009
• February 2009
• March 2009
• April 2009
• June 2009
• August 2009
• September 2009
• October 2009
• December 2009
• January 2010
• February 2010
• March 2010