Mizzou researchers create a tool that makes Twitter more powerful

Mizzou assistant professor Sean Goggins (left) and doctoral student Ian Graves developed software that measures the context of words in Twitter. (Photo courtesy of the MU News Bureau)

Mizzou assistant professor Sean Goggins (left) and doctoral student Ian Graves developed software that measures the context of words used in Twitter. (Photo courtesy of the MU News Bureau)

Twitter already is a powerful news aggregator and microblogging platform. Now, two University of Missouri researchers think they know how to improve it.

Their thinking stems from new software the pair developed that they say considers the context of tweets, not just the quantity. At present, a topic is popular or “trending” on Twitter if there are a high number of related keywords and hashtags that are associated with it.

But the software, developed by Mizzou assistant professor Sean Goggins and doctoral student Ian Graves, can be programmed to pick out words and analyze their placement within tweets.

Goggins and Graves said they tested their concept on a flurry of tweets from the Super Bowl and World Series and assigned tags to words they predicted would be common in the two broad conversations. The software scrutinized where the words were located in each tweet, thus giving the researchers notions on the words’ contextual importance and allowing them to see how conversations evolved.

“When analyzing tweets that are connected to an action or an event, looking for specific words at the beginning of the tweets gives us a better indication of what is occurring, rather than only looking at hashtags,” Goggins said in a Mizzou news release.

In tracking word placement, the researchers were able to determine the nuance attached to each Twitter discussion. They could discern the action on the ball field between pitches and on the gridiron between plays.

“The program uses a computational approach to seek out not only a spike in hashtags or words, but also what’s really happening on a micro-level,” Graves said. “By looking for low-volume, localized tweets, we gleaned intelligence that stood apart from the clutter and noise” associated with each event.

Goggins and Graves believe their software will help make Twitter more effective for monitoring community safety and tracking disaster relief, and improve understanding of cause and effect in major events such as the bombings at the Boston Marathon and the protests in Ferguson.

Although less than 5 percent of Twitter traffic is actual news, much of the dialog that drives retweets and hashtags relates to newsworthy events.

Goggins teaches in the School of Information Science and Learning Technologies at Mizzou. Graves is a student in the Computer Science and IT Department at Mizzou’s College of Engineering. Nora McDonald, a graduate student at Drexel University, contributed to the study, which appears in the journal New Media and Society and was funded by a grant by the National Science Foundation.

KMOV: Setting a bad example

KMOV logoFrom the first day of my course in basic journalism at Lindenwood University last semester, I hammered into my students’ heads the importance of accuracy in reporting.

It was an essential part of my lectures, my assignments and my grading system ― so much so that the students were ordered to supply me with contact information for the sources in each of their stories. Any detail they cited had to be referenced, and that reference had to have an email address or phone number attached for me to verify.

If doubt trumped veracity, their grades suffered. Heavily.

“Accuracy is at the core of your credibility,” I said and posted in a PowerPoint presentation. “Subtract that and you’re less of a journalist, less of a professional.”

I wonder now if I should extend a formal invitation to the reporting staff of KMOV-TV to take my course. Because recent events involving the CBS affiliate have put into question its appreciation of accuracy in reporting.

The first event, profiled here last month, involved former KMOV news anchor Larry Conners, who stirred protest and scuttled his job by alleging via Facebook that the Internal Revenue Service was harassing him because of an interview in April 2012 with President Barack Obama that Conners believed put the president on the defensive.

Conners admitted in the Facebook post he had no proof but neglected to mention that his issues with the IRS went back at least four years before the interview. The station soon fired Conners, accusing him of harboring bias and dragging KMOV’s name through his speculation.

Conners insists he was just doing his job. He’s busy now however leveling another accusation, having filed a discrimination suit against KMOV on a peripheral matter.

Event No. 2 blew through Twitter on Friday evening as tornadic winds bounded between St. Charles and St. Louis, followed by a flood of tweets saying KMOV had reported on television around 8:30 that a “mass casualty” event involved a storm-wrecked hotel in the storm’s path.

The phrase echoed ominously across social media as the Twitterverse awaited a citable source from KMOV confirming the destruction. No other news provider offered similar reports or alternate confirmation, and at least one wondered openly where KMOV was getting its information. Meanwhile, social media watchers said KMOV kept repeating the frightening words on the air.

KMOV began backing away from its initial televised report about 30 minutes later, but not before changing the location of destruction and leaving St. Louis County authorities and representatives of the hotels that were named to assuage fears via their own social media. By then, the storm seemed secondary on Twitter to KMOV’s own hasty, alarming damage assessment.

Twitterers continued hurling brickbats at the station well into the next day. Even social media maven Andy Carvin of National Public Radio weighed in.

“Yet another twitter rumor spread because of poor initial reporting by mainstream media,” he tweeted Saturday.

Though KMOV never explained itself, the initial report of mass storm casualties was thought to be inspired by a vague understanding of dialog emanating from a police scanner ― historically, an unreliable source for factual information. You would think KMOV already knew this.

“Police reporters depend on sources in the department and on their knowledge of police procedure for their stories,” wrote educator Melvin Mencher in his college text “News Reporting and Writing,” now in its 12th edition. He later added that, “Sins of omission occur when the journalist fails to act in situations in which revelation is required. … More often, the omission is the result of laziness or ignorance.”

A group effort by Fred Fedler, John R. Bender, Lucinda Davenport and Michael R. Drager titled “Reporting for the Media” reiterates this point.

“If reporters lack some information, they should consult their sources again,” the authors wrote. “Reporters should never guess or make assumptions about the facts. … Conscientious news organizations check their stories’ accuracy.”

And former newspaper editor Tim Harrower devotes a portion of his own popular textbook, “Inside Reporting: A Practical Guide to the Craft of Journalism,” to covering accidents and disasters. Intrinsic to this kind of journalism: confirming before reporting.

“No matter how useful the Internet may be, it’s no substitute for reality ― for real discussions with real human beings,” Harrower wrote.

All of these texts have been around at least 10 years and are staples of journalism education. It’s hard to imagine that anyone at KMOV who studied journalism hasn’t read one, or read one like them.

But if they haven’t, the Boston Marathon bombings offered a contemporary lesson as twitterers latched onto police scanner reports of the manhunt for the bombing suspects, and ensuing urban lockdown, and quoted like gospel every snatch of detail and garbled bit of dialog.

“Any reporter who was trained in an honest-to-goodness newsroom knows this much: The police scanner is a blunt instrument, not a source of solid facts,” wrote Curt Woodward in Cognoscenti, assessing the manhunt’s impact. “… The stuff being said over those airwaves is definitely real. But it isn’t necessarily true.

“Civilians can be forgiven for not knowing this. But professional journalists? Yikes,” he continued. “If you care about your audience, you don’t report what you hear coming over the scanner, without confirming it first.”

True, the proximity of Conners’ situation and KMOV’s storm reporting were too close to be anything but coincidence. Still, the dual social media failures imply a pattern of behavior, a misunderstanding about the importance, relevance and sensitivity of social media usage in news reporting.

So, I welcome staffers at KMOV to sit in on my classes next semester, particularly the one where we discuss how best to use tools such as Facebook and Twitter, and review how the station can start setting a better example. Or, I can lend them the textbooks; I have all three.

Or, maybe I’ll just have one of my students explain it to them.