Why Facebook’s Trending Algorithm Poses More Problems on Human Bias

Sep 02, 2016 05:16 AM EDT | Rubi Valdez

Facebook Trending published fictitious story on Fox News anchor Megyn Kelly

Fox News anchor Megyn Kelly was recently involved in a viral story published online saying she was fired at the network due to her campaigning of presidential candidate Hillary Clinton. Kelly denied the rumor and Facebook has since taken down the post.
(Photo : Chip Somodevilla / Staff)

Last Friday, Facebook fired its team of human curators for the Trending platform due to reported manual insertion of topics. The company's statement read that the move is intended to reduce bias and "make the product more automated."

Facebook said stories on Trending should not necessarily be treated as straight news because it represents "algorithmically detected topics that reflect real-world events." Unfortunately, it backlashed when netizens reacted negatively to a false news story about Fox News anchor Megyn Kelly being fired due to her political inclinations.

Byron Galbraith, chief data scientist at Talla, said Facebook's Trending algorithm cannot resolve problems brought by human bias. "It's an interesting challenge because it speaks to the problem of our algorithms," he said. "The data that derives the algorithm is inherently biased."

Trending engineers can only supervise how the algorithm works but not the actual content of the populated stories. Facebook said the people in the team work for guidance only and they have no editorial privilege.

Basically, "Topics related to sex, pornography, nudity, graphic violence, etc. Stories that can be perceived as R-rated or worse should be tagged risqué," cited Quartz. Facebook populates trending topics based on Mentions, a key term that is talked about by a number of people online. Demographic attributes like gender, race, and location influence how a topic is seen as trending or viral.

CNBC reported that Facebook needs to deploy humans to ensure quality in determining if a story is fictitious or not. Matt Lang, a senior digital strategist at Rain, said there is still a long way to go to clean up the algorithm and significantly reduce bias.

However, Facebook can implement a precautionary measure to distance itself from an accusation of bias. Galbraith said suggesting articles of conservative and liberal tone should be able to strike a balance.

© 2017 Auto World News, All rights reserved. Do not reproduce without permission.
Get the Most Popular Autoworld Stories in a Weekly Newsletter
`

Join the Conversation

Real Time Analytics