Earlier this year, several producers at CNN received an alert from a new digital news gathering tool that they were testing: A teenager in Los Angeles had posted a tweet saying that the pop music heartthrob Justin Bieber had been arrested.
CNN’s Los Angeles bureau followed up on the tip by calling the appropriate local precinct. The police confirmed that Mr. Bieber had just been arrested on burglary charges, but wanted to know how the cable news network could possibly have known that.
The answer is Dataminr, a software tool designed specifically to analyze billions of Twitter postings for patterns that could indicate breaking news. On Tuesday, Dataminr becomes commercially available to news organizations, some of which — like CNN and Gannett — have already been testing the software.
“On the whole this it is one of the best technical tools we have,” said Richard T. Griffiths, senior editorial director at CNN. “Besides our local affiliates, it is probably our best sources of story leads, and you live or die by your leads.”
Dataminr was started in 2009 as an information-discovery tool for Wall Street traders and money managers looking to gain an investment edge as well as for public-sector institutions. But the company has broadened its services as news organizations look for ways to analyze the vast amounts of information on social media like Twitter and YouTube for the next big thing, whether it be news or a hot new band.
Groups of businesses have formed around mining social media for its trove of information. Storyful, which scours sites like Twitter and Instagram for breaking news, was purchased by News Corporation last December. Vocativ, which uses algorithms to search YouTube for emerging video, announced a partnership with NBC. And 300 is a company that plumbs the depths of Twitter’s data trove for information about musical acts.
Ted Bailey, the founder and chief executive of Dataminr, said his service allows journalists to customize their own settings and focus on their specific areas of interest. The software searches for patterns in Twitter that would indicate a noteworthy event is taking place, like a sudden increase in tweets in a narrow geographic region over an unusual time frame.
The Dataminr software, which was developed in conjunction with Twitter, searches the 500 million or so daily postings to Twitter, Mr. Bailey said. It then culls the information to just a few urgent alerts a days, on average. “What makes Dataminr so effective is how infrequently it alerts you,” he said, “without having you miss anything important.”
Beryl Love, executive editor of the USA Today national news network, which serves the flagship paper as well as dozens of others across the country, said his company was paying for Dataminr’s service because it gave a deeper Twitter search than any other tool he knew of, including putting tweets in a timeline or searching only for those with attached images.
“It is now how we do surveillance on a daily basis,” he said.