A traditional Ork Poster for Seattle

This is pretty cool. Microsoft researcher Andrés Monroy-Hernández, an expert in social media analysis, has come up with an interesting way to view Seattle. (One we’ve never quite seen before).

Think of it as the citywide Tweet map — in the style of the popular Ork Posters.  The map was created using Twitter data over a two month period, so for example you can see that words such as “coffee” and “weekend” are associated with Capitol Hill or “Seattle PD” and “apple” are associated with the U. District.

“The (map) is definitely biased towards volume. So the loudest is the most visible, which is part of the provocation,” says Monroy-Hernández in a Tweet to describing the map.

I’ve got to say that I am not too wild about the idea of living in an area where the most prominent words are “response armed.”

Ugh…

Maybe I’ll move over to the Roosevelt area where they seem to love “beer” and “toys.” There are some oddities on the map, like the fact that “wallingford” is one of the the prominent words in the heart of the Ballard neighborhood. I am also not quite sure why Seward Park has been labeled “broken,” or why “salt” appears to be such an important issue in the Lake City area.

There are also a few Twitter handles that make an appearance on the map like @jcrossover — the handle for former Rainier Beach High School basketball star Jamal Crawford — and @jseattle — the handle for Capitol Hill Seattle blogger Justin Carder.

UPDATEMonroy-Hernández tells us that map came about as part of the Maker Fest at Microsoft Research, since he wanted to pursue the idea of “data-driven art” that he could imagine hanging on his walls.

“As I recently moved to Seattle, I have both empty walls and an interest in art that represents my new city,” he said.

Here’s more on how the Tweet map works:

Since it’s volume driven, the neighborhoods with low activity are over-represented by tweets from the police department. So it was partly provocative, to say: ‘hey, not everyone is tweeting at the same rate.’ We extracted all the tweets of people whose profile says they live in Seattle in a period of two months. We got a list of each neighborhood in Seattle, then we identified tweets by neighborhood using specific keywords — for example “cap hill,” “capitol hill,” even misspellings like “capital hill.” Then once we had a data set for each neighborhood, we then combined neighborhoods that were very close to each other, such as Montlake and Capitol Hill, so that the visuals were easier to read.

Take a look and let us know what you think. What word would you like most associated with your neighborhood? Does this accurately represent your neighborhood?

Click on map for larger version

Latest News

Comments

  • Guest

    Those keywords need context to make sense. I assume FREE in West Seattle is because of the battle over free parking spaces being chewed up by fees lately? There ain’t anything else free in West Seattle.

  • Nathan Stephens

    When do they create an app to do this for everyone? This is great!

  • http://twitter.com/protectedstatic protected static

    “Salt” in “Lake City” is an algorithm problem: it’s pulling in Salt Lake City, which to my mind suggests issues with the whole map.

  • http://twitter.com/Ell79 Elliot McDonell

    Yes, context. Salt/Lake City would probably be tweets about SLC, UT, from folks in Seattle. Otherwise a cool project!

  • jorge

    The denser the neighborhood, the smaller the size which lends a higher visual weight to areas which have fewer tweets. Methinks the representation is awry. ;)

  • http://superkinz.com Andrew Kinzer

    I’d say “morning” represents Ballard well. We have one every day!

  • http://www.facebook.com/vqnguyen2 Viet Q. Nguyen

    The giant “HELP” in capital letters along the southern part of Aurora near Denny accurately describes this neighborhood.

  • Anony

    There’s also “hoyas” down on Harbor Island, which seems to indicate a conflation of our Georgetown with the one in DC. Given that many Seattle neighborhoods borrow or share their names with other, more famous locales—Ravenna, Wallingford, South Park, Capitol Hill, Fremont, Beacon Hill, etc—this makes the whole exercise seem pretty useless. Perhaps it would’ve been better to use only geotagged tweets, even though that would’ve further limited the pool. As it is, those scary SPD tweets are probably the only ones we can be certain actually came from the areas in question.

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.