The next step in the future of journalism could be led by a Seattle startup. Today, media giant Hearst announced it has started a new website called LMK (Let Me Know), a news aggregator that pulls in Web feeds from sources like the Associated Press and Getty Images, and automatically creates topic pages for individual celebrities, sports teams, companies, and the like. The content filtering technology behind the site comes courtesy of a partnership with Seattle-based Evri, a startup founded by Paul Allen’s Vulcan Capital.
It’s a big deal for Evri, which has been using semantic analysis and natural language processing to find meaningful connections between entities on the Web. Its technology has led to partnerships with The Washington Post and The Times of London, in which Evri’s widget suggests related articles for readers. But the Hearst deal is more significant, because Evri is actually powering the site.
Will Hunsinger, Evri’s chief executive, says the deal was done about a month ago. He declined to give financial details, but did talk about how the partnership works. Hearst subscribes to Evri’s Web platform, which ingests more than 30,000 sources of content (news, blogs, tweets) and applies natural language processing to produce a massive “semantic index.” When the LMK site wants information about the Florida Gators, say—one of the site’s first focuses is college football—Evri’s software filters stories about the Gators by their timeliness, relevancy, and credibility of sources. It then delivers the primary links and snippets of text and photos to LMK.
“What’s significant is we feel this is a beachhead, or flagship demonstration, of how you can use semantic technology to deliver what appears to be an editorial experience, algorithmically,” Hunsinger says. The benefits of this are twofold. It’s a very fast way of getting news up, and it’s cheap—you don’t need a big editorial staff to pick and choose each story.
Of course, its value all depends on how good the feeds and filtering are. Hunsinger says Evri’s semantic analysis could make news aggregators far more powerful than sites like Google News. That’s because Evri’s software understands that Manny Ramirez is a baseball player associated with steroids, for example—so you can filter content by asking for all stories about baseball players, or stories about players using banned substances, and Ramirez will show up. Most aggregators, including Google, are based on keywords, so you’d usually have to type in “Manny Ramirez” to get stories about him to appear.
Beyond LMK, technology like Evri’s could potentially make the whole editorial side of the newspaper industry more efficient, Hunsinger says. “You could deliver all stories around a particular topic—the entire story behind the story,” he says. That might let reporters and editors spend more time on things like developing relationships with sources and digging up entirely new stories. Examples of aggregation topics might be Seattle sports, or San Francisco politics.
I asked Hunsinger whether being a huge sports fan had anything to do with getting the LMK deal done. He insists it was just “a total bonus.” But he added that his stress level will be going up in about a month, when his alma mater, the Georgetown Hoyas, start their basketball season.
