How SmartNews leverages AI to drive relevancy for readers: Rich Jaroslovsky

By Christi Warren

SmartNews, a news-discovery app, was created to provide efficient and fast, well-curated news for readers who are on-the-go.

At the heart of what makes it so successful is machine learning, which provides an understanding of what news matters most to which users.

Using data, the algorithm is able to surface quality news stories from unexpected sources — outlets a certain reader might not always go to for their news, said Rich Jaroslovksy, vice president of content and chief journalist for SmartNews. Jaroslovsky spoke at the 10^th Data & AI for Media Week at Microsoft in San Francisco on May 2.

Unlike other news apps, SmartNews is optimized for discovery and diversity over personalization, he said.

Machine learning allows SmartNews to evaluate many more stories in a very short period of time than any team of humans ever possibly could.

“We’re able to key on the attributes of that content that signal that it is high-quality content, and we’re able to scale it,” Jarosolovsky said. “We’re using machine learning to analyze word-by-word the contents of that article so we can classify it, so we can group stories about the same topics together so that we’re able to weed out duplicates. We’re categorizing, ranking the stories by importance, and all of this is happening in real time.”