Content sourced from social media can be the most powerful and engaging way to tell a story. Leading journalists are mining social platforms to diversify their network of sources and discover unique content from the scene of earthquakes, hurricanes, and other natural disasters. These updates from the scene can also surface new, compelling angles for reporters covering the human interest angle of these breaking stories.

In a live Webinar on July 25, Meghan Plambeck, director of Dataminr For News, presented on some of the key resources and practices for using social media to cover natural disaster reporting. Dataminr is an early alert warning system for disasters, that integrates with mobile and a host of social media platforms.

The Webinar covered why social media and user-generated content (USG) are impactful for storytelling during disasters, as well as the challenges of social news gathering, with examples from media coverage of sandstorms, cyclones, volcanoes, hurricanes, and other events.

“Content from social media can be the most powerful and engaging way to tell the story,” Plambeck said. “Content that’s shared in real time from people who are on the ground, sharing what’s happening around them during the event is very powerful. Those social posts that people are experiencing as things are unfolding are really powerful.”

Some of the events covered in this manner can reveal stories and images that are difficult for the audience to find in searches or from traditional reporting. Dataminr can uncover these types of stories without use of certain keywords — often just moments after they happen.

Dataminr allows journalists to source reports, images, and videos from the scene of natural disasters.
Dataminr allows journalists to source reports, images, and videos from the scene of natural disasters.

“Dataminr is language agnostic,” Plambeck said. “We are optimised for Twitter, but if you weren’t searching in native language, you would not be able to surface this kind of content on your own.”

Benefits of social media reporting

Plambeck highlighted four benefits of reporting on natural disasters using social media content:

  1. Speed is an advantage — news from disasters often is first reported on social ahead of the wire services.
  2. Social provides immediate access to a variety of sources: eyewitnesses, emergency responders, government and public safety officials, and other reporters/media outlets.
  3. Visiting an area following a major catastrophe can put the journalist in danger or the area may be inaccessible; in those cases, social can allow them to report until it’s safe to access the site.
  4. Photos and video will re-circulate on social networks from previous similar events, so it’s critical to confirm it’s accurate before re-sharing or using in story coverage. Reference firstdraftnews.org/learn for training on social media verification.

“In the absence of actually having boots on the ground in places where these natural disasters are unfolding, social can be incredibly helpful in allowing you to have eyes and ears on the ground in places where you may not have physical reporters,” Plambeck said.

Challenges of social media reporting

One of the major challenges to this is volume. With 500 million tweets sent daily, how to you find the signal in the noise?

“What we do is we apply our technology to every publicly available tweet,” Plambeck said. “How do you find the most important, most relevant information that’s happening?”

That can be difficult to do because hashtags don’t always reveal them, and reporters don’t know who will post breaking information. It might be someone a reporter follows on Twitter, but most often it’s not.

“It’s some random person who just happens to be standing in front of a volcano when it blows. What Dataminr is meant to do is to find those pieces of information that are relevant to what you’re covering, and then deliver them to you proactively.”

Another challenge is language — if breaking news is happening in a place where your organisation doesn’t speak that language natively, finding and reporting on that news is difficult.

Though Dataminr is optimised for Twitter, it also surfaces cross-posted content from other platforms.

Examples from various natural disasters

In a sandstorm in a remote part of India, a major newspaper was able to cover the story not because they could send a reporter there, but because they were able to use Dataminr to find a group of tourists who were at the location participating in a rickshaw race. The media organisation was able to connect with one of those participants to build out a fascinating and well-reported story on the event.

A group participating in a rickshaw race in remote India during a sandstorm became a great source for covering the disaster.
A group participating in a rickshaw race in remote India during a sandstorm became a great source for covering the disaster.

During Cyclone Debbie near Australia, the “Sharknado” feature story first surfaced on Twitter. “These can also give your newsrooms early ideas for feature content,” Plambeck explained. During the cyclone, a story surfaced about a bull shark that had been picked up in the storm and thrown out of the water.

“Newsrooms that received that alert, they reported this as a real-life sharknado. Not necessarily discreet breaking news, but the kind of thing that adds an additional angle and additional context that can really resonate with the audience.”

As Volcan de Fuego erupted in Guatemala recently, social coverage was crucial for the video updates that they circulated. “This may be the closest you’re going to get to this particular event,” Plambeck said. “These videos will be super compelling but difficult to surface.”

Videos sourced through Dataminr provided compelling news coverage of the Volcan de Fuego eruption in Guatemala.
Videos sourced through Dataminr provided compelling news coverage of the Volcan de Fuego eruption in Guatemala.

Between language barriers in searching and the huge amount of information out there when such an event strikes, it’s hard to find just the right kind of content — and to continue finding more. “What Dataminr allows you to do is to opt-in to what we call ‘track story,’” Plambeck explained. The system will then keep you updated on an ongoing basis for new information.

Reminders for using social to report on natural disasters

Plambeck went over several key things that news companies need to keep in mind when using social media to cover disasters:

  • Always consider the safety of your social media source, who may be posting from a potentially life-threatening situation.
  • Photos and video will re-circulate on social networks from previous similar events, so it’s critical to confirm it’s accurate before resharing or using in story coverage. Reference FirstDraftNews.org for guides and training on social media verification.
  • Be aware of the psychological effects of reporting on catastrophic events and what companies can do to protect their reporters. Resources are available from the Dart Center for Journalism and Trauma

Q&A

INMA: It looks like you’re analysing images to detect disasters, not just text. How do you do that?

Plambeck: It’s pattern recognition really, not just text. So we’re looking for anomalies in the data based on previous disasters. Our data goes back to every available tweet since the beginning of Twitter.

INMA: How much data do you find is geo-tagged?

Plambeck: I can only speak for Twitter, but only about 2% of people geo-tag themselves on Twitter. However, Dataminr has some features that allow prediction of locations and zero in on special geographical areas.

INMA: Can Dataminr find content from social media platforms in China?

Plambeck: Obviously the system can only source alerts from tweets, so the answer to that question broadly is not well because there are very few tweets in China. Of course if something else is referencing something in China, yes Dataminr would find that.

INMA: How does Dataminr determine whether the tweet is fake?

Plambeck: What you’ll never see from Dataminr is a flat-out wrong piece of information. But what’s important to realise is that it’s a prediction tool. Using data from the past, it’s making predictions going forward about the type of news event and what will be covered.

Typically we do see it happen where an event may seem like a huge thing at the outset, but may not turn out to be that big of an event. (For example, early reports of shots that turn out not to be what it was initially thought). There’s nothing more frustrating than following a false lead, but most journalists would rather know that something might be happening and follow it to find out rather than miss it.

INMA: How do you validate video or a picture posted by a social media user exclusively?

Plambeck: Nothing you get from Dataminr is validated, that’s important to know. We are sourcing and giving you the information, but it’s the journalist’s job to confirm that information.