Facebook has rolled out a new update that seeks to cut down on spammy ads on users’ news feeds.
The announcement came just over a month after the social media site clamped down on fake news and disinformation.
“With this update, we reviewed hundreds of thousands of web pages linked to from Facebook to identify those that contain little substantive content and have a large number of disruptive, shocking or malicious ads,” the blog post, written by Jiun-Ren Lin and Shengbo Guo, said.
This is not a new policy, of course, as engineers at the social media site have been trying to weed out low quality or “disruptive” content from its pages since last year. Tabloid-style headlines, deceptive ads, sexual images, and shocking content are either pushed back down in the feeds or erased altogether.
— craig daniels (@organichat) May 10, 2017
Facebook utilized machine learning to study the blueprint and models used by spammy ads. With the new algorithm in hand, its AI systems then comb through the billions of posts to find similar patterns.
One of the reasons for Facebook’s aggressive approach toward spammy ads, fake news, and disruptive content was the blowback that Mark Zuckerberg received from critics who accused the social media company of influencing the U.S. elections last November.
Zuckerberg initially hedged, but immediately introduced measures to insulate the platform from being a harbinger of fake news and disinformation. In February, the billionaire released a manifesto where he reiterated the company’s original vision of building a community to bring the world closer together.
In this manifesto, he revealed that they were working on machine learning to flag videos and photos that are deemed to be controversial. The Facebook team will then review the content before allowing it to be seen on users’ news feeds. “Today I want to focus on the most important question of all: Are we building the world we all want?” Zuckerberg wrote.
Greg Marra, the company’s product manager for news feed, said the system is not perfect, but the initial results have been encouraging. Below are some of the criteria Facebook’s machines are looking at to categorize a specific content as low-quality or disruptive:
- Does it have a “significant amount of original content?”
- Are there pop-up ads on the landing pages?
- Are the ads shocking or sexual?
Marra said their algorithms will skip over some ads if they are “high-quality” enough that people wouldn’t mind seeing them when they do click on a sponsored content in their news feeds.