fbpx
News Hub

Facebook to introduce explainability to ad and news feed algorithms

Written by Mon 1 Apr 2019

New tool will enable users to see how Facebook ranks posts based on user habits

Facebook is to begin explaining to users why they are seeing certain posts on their news feed in an attempt to make the platform easier to understand.

The social network has announced it is to start offering users a “why am I seeing this post?” tab, which they can click on to see how Facebook’s algorithms decided it was a relevant post.

The site has also announced an update to its advertising tools which will enable users to more transparently see how they were targeted by an advertiser on the site.

Both tools will be accessible from the drop-down menu in the top right-hand corner of any post or advert on the news feed of the social media platform.

The new news feed tab will show a user’s recent activity that helped the site’s algorithms reach its decision, including any links or interactions with the posting account and if they have a history of reading posts from that account.

Explainability

More and more enterprises are relying on machine learning techniques to build models that can predict and classify certain phenomena, such as images of cancerous cells or credit risk.

As these models become more complex, it becomes increasingly difficult to explain why they make the decisions they do. This can result in individuals being declined loans without clear information as to why they are not approved.

To combat this problem, there is a growing consensus in the machine learning and political communities that techniques need to be developed that allow firms to explain why their algorithms arrive at certain conclusions.

Facebook said the new feature was designed to help users “better understand and more easily control what you see from friends, pages and groups in your news feed”.

The firm also announced it was updating its similar “why am I seeing this ad?” tool so that users can see how advertisers can use existing customer lists to show adverts to people on the platform.

Advertisers are able to upload lists of existing customer information – such as email addresses or phone numbers – to help target their advertising.

Facebook said this process takes place without revealing any identifiable information but added the new tool would more clearly show how this process took place.

Written by Mon 1 Apr 2019

Tags:

AI Facebook machine learning social media
Send us a correction Send us a news tip