If you've been following the Post's political coverage, you've probably come across @MentionMachine. The tool, which tracks mentions of the 2012 presidential candidates on Twitter and in the media generally, launched Tuesday and is the brainchild of Executive Producer Cory Haik.
I spoke with Haik Friday morning to discuss how @MentionMachine came to be and the impact it might have on election reporting.
"Well, we'd been talking about Twitter and the elections since I'd walked in the door of the Post," said Haik, who started a little over a year ago as deputy editor of the Post's Universal News Desk. She has since become Executive Producer for News Innovation and Strategic Projects.
Concepts can either fizzle or thrive depending on available resources. In this case, Haik had already chosen a name. "I feel like once you give something a name, it becomes a thing," she said. "I squatted on the Twitter account long before it was even an official project, because I thought, 'This thing's gonna' work.'"
After a series of memos, meetings, and presentations on a wash-rinse-and-repeat cycle, Haik started to bring newsroom resources to bear on the project, teaming up with Washington Post National Digital Editor Amanda Zamora, Developers Jesse Foltz and Sean McBride, Designer Katie Parker and PostPolitics Social Media Producer Natalie Jennings.
What resulted is a tool that is more complex than the otherwise simple menu bar would indicate. Click on a candidate and you can see how often they are mentioned on Twitter relative to other Republican primary candidates and President Obama.
But wait, there's more.
"We keep talking about Twitter, but we're also counting Trove," said Haik of the Post-acquired company that brings in more than 10,000 publications, targeting them to users' preferences. "So, that's an awesome win there."
Jennings reports on the @MentionMachine data regularly, mining it to help inform the Post's journalism. "The daily data is proving itself," said Haik, "and doing that reporting is a big part of the machine. It's not just the numbers."
The @MentionMachine has a profanity filter, but tweets with profanity are counted among the candidates' mentions. Try to get any more specific than that, and you run into what Haik calls "a special sauce" -- the programming muscle that powers @MentionMachine. Haik says she and others refer to it as a combination of Twitter, her own unique innovations, and the execution engine Kangaroo.
Registered voters are not given preference, since @MentionMachine does not target those in the U.S. or provide U.S. geotagged tweets any additional weight. @MentionMachine does, however, feature "Top Tweets" -- or tweets that are considered to be driving the conversation about a particular candidate at the given moment. "We have a very unique formula algorithm for top tweets, it's not just retweets," said Haik, emphasizing that there are safeguards in place against individuals trying to game the system.
"We're making sure that our top tweets are tweets that are driving the conversation," she continued, "So, the point is we're trying to show people exactly where people are spiking and when."
The Post is not the only organization trying to mine what has come to be known as "big data" when it comes to the 2012 election. CNN debuted a similar Twitter tracking tool on Iowa caucus night, and other Web sites are drawing from the Twitter API data pipeline as well. Regardless of what others are doing, Haik said she's confident that @MentionMachine has carved a unique place for itself as a transparent tool able to provide journalistic value. "We tried to make the machine programatically do some journalism for us," she said.
While @MentionMachine shows candidate mentions, Haik is often asked about sentiment data -- or whether a tweet or media mention can be classified as falling into one or more emotional categories, such as sarcasm, anger, sadness, and joy. "Sentiment data's not there yet," said Haik, "I don't feel like you have to have sentiment data to get a good read on where the
conversation is on a person."
The @MentionMachine is by no means static. It is continually tweaked and improved. "I'm calling it a very mature prototype," said Haik. She provided a preview of one impending change: "What you're going to see next week is when you hover over that spike tweet, you're going to get that exact top tweet for that moment."
Read more about @MentionMachine and explore the data.
