Science information channels used to be pretty limited. Scientists would publish their findings in obscure journals, and only other researchers would ever see them.
A new generation of scientists armed with new technology tools, have led to some big changes.
“I’ve grown up with this idea that science should be open. There are a lot of ways people can share information back and forth through serendipitous means on the Web,” says Erin Robinson, a graduate engineering student at Washington University in St. Louis, Mo.
Just as businesses now make some decisions by monitoring social media chatter, and news organizations rely on “citizen journalists” for a first look at breaking news, scientists are letting “civilians” help them with research.
Hence the birth of “Air Twitter,” a project that examines air quality.
With some filtering and some analysis, scientists are able to take mundane texts and tweets and turn them into enormous amounts of accurate information.
Air, like weather, is something everyone talks about.
“Air quality is special; it affects daily life. People notice it; they are exposed to it,” Robinson says.
And, a few nanoseconds after they notice the smog or the haze or the dust storm, they pull out a phone and share their observations with the world.
Scientists like Robinson realized the information was already being shared — they just needed a way to harness it.
Earth science for everyone
Air Twitter is a project of ESIP, the Federation of Earth Science Information Partners. It’s a network of 121 organizations, from NASA, NOAA and USGS data centers, to supercomputing facilities, nonprofits and businesses. Its work is aimed at making earth science information available for everyone.
Robinson says this was a group already receptive to the idea of using the Web in new ways. She’s put the project into operation with programmer Ed Fialkowsky.
“The initial hypothesis was to see if we could identify air quality events from the social media. What we found was, we could,” says Robinson.
So, how do they take messages like “OMG, I can hardly breathe today!” or “The wildfire is coming over the ridge behind my house!” and turn them into useful scientific data?
This social media listening tool first harvests content, from Twitter, flickr, YouTube, and delicious, that have keywords like “air quality,” “fire” and “smoke.”
Those entries are aggregated, and filtered. Then relevant air quality content is re-tweeted, to the ESIP Air Quality Working Group. (Followers include California Gov. Arnold Schwarzenegger and London Mayor Boris Johnson.)
The scientists then give back information on a Wiki page, which can be important, even life-saving, in an emergency.
“We develop a workspace that links together the people who are tweeting and posting pictures about the event, and the news stories, and we combine that with the science data, from the satellite and the surface monitors,” says Robinson.
Trial by wildfire
Shortly after the project started, Robinson put together a Wiki page to follow a California wildfire.
“It turned out that this event just got bigger and bigger; southern California was on fire for a month,” she says.
The information was fast, and accurate. People affected by the fires could find weather forecasts, hazard maps, smoke models, photos and videos.
“Using the social media, you can identify the event probably 24 hours quicker than you can with the rest of the science data. You don’t have surface observations normally for 24-36 hours, and the satellites are usually about a day behind,” says Robinson.
“The idea at the international level is that there have to be systems in place for free and open exchange of data, because the problems that we are facing, like the oil spill, climate change, issues around food, are all so serious, and so urgent,” says Robinson.
Other big issues may also benefit from social media listening, from disaster management to public health.
Just as millions of people tweet about air quality, they also tweet about their health.
Twitter tracks flu outbreaks, too
In a different study, computer scientist Aron Culotta tracked 500 million Twitter messages to track flu outbreaks. His work at Southeastern Louisiana University turned out to have a 95 percent correlation with the national health statistics collected by the Centers for Disease Control and Prevention (CDC). And it was much cheaper and faster.
“It was very surprising to me that the correlation is so strong,” says Culotta.
He and his students produced flu estimates daily, while the CDC information lagged by a week or two. His work, funded by the Louisiana Board of Regents, isn’t aimed at replacing what the CDC does, but possibly improving it.
Keeping track of a fast-moving disease could help individuals and the health care community prepare for an outbreak. And Culotta says both business and government now “get” that this is important work.
“Marketing, the government, everyone is starting to catch on that this is a valuable source of, I like to think of this as ‘passive polling.’ You don’t get on the phone and ask people how they are feeling — they’re telling everybody on a daily basis!” he says.
“We just need to write the algorithms that can understand those messages,” says Culotta.