How Flipboard Sorts Through Hundreds of Thousands of Articles Each Day

How Flipboard Sorts Through Hundreds of Thousands of Articles Each Day

AI drives recommendations, but humans still play a role

How will artificial intelligence (AI) affect the business of digital publishing?

A number of publishers first focused on automating the writing process, with bots creating stories based on structured data such as box scores and financial reports.

But according to a recent survey of publishers worldwide from the Reuters Institute for the Study of Journalism at the University of Oxford, content recommendation is now the single most common use of AI in publishing.

Flipboard, a news aggregation platform, is an extreme testing ground for AI recommendations, given that it features as many as 300,000 articles each day. eMarketer spoke with Mike Cora, Flipboard's engineering manager for data products, about how the platform uses AI to make suggestions to its users.

Of the 150 people who work for Flipboard, about 40 are focused on tech, engineering and data science, which are the teams responsible for creating and monitoring AI tools. Those tools scan for contextual clues to tag articles by topic and keyword, Cora said.

The platform’s AI features also weed out identical articles, and block spam domains that attempt to spoof legitimate websites.

AI does not manage the entire recommendation process, however. Flipboard has a team of 20 curators who manually select articles to recommend to its readers, who the company says number more than 100 million per month.

Like other companies, Flipboard built its AI platform on the work of academic researchers. But, Cora noted, creating the proper code to make theoretical applications a reality was not a trivial assignment.

An added complication is foreign language processing. Flipboard aggregates content from sources in multiple markets, and because different languages have different grammatical structures and expressions, the AI tools have to be trained to detect different signals for different languages. It took almost a year of research and development—and then another six months for all the code to be put in production—before Flipboard got its platform's contextual detection capabilities up to par across multiple languages, according to Cora.

Like Flipboard, other publishers around the world are tapping into AI to improve their content recommendations. Some 60% of the 184 publishers in the Reuters Institute poll said they are using AI to improve content recommendations. Significant numbers were also using it to automate workflows and to target ads. (Flipboard, too, uses AI for ad targeting.)

The recommendation challenge is not limited to publishers, of course, but is common to media companies of almost every type. Platforms like Spotify and Netflix use AI and machine learning algorithms to drive their content recommendations. And publishers such as CBS and Rodale use AI vendors, like Iris.TV, to determine which video playlists they should serve users.

Share this Article

Similar Stories