Why LinkedIn’s Feed Algorithm Needs a Revamp

Spam can make or mar our social media experience. A study by Nexgate found that spam on social networking sites was growing at a faster rate than comments. To this end, last week, LinkedIn released a blog titled, ‘Viral spam content detection at LinkedIn’, where the platform claimed that it uses proactive and reactive defences to combat spam and malicious content on its platform.  The platform’s proactive defences consist of two types of classifiers that analyse specific spam categories and content types, using deep neural networks trained with TensorFlow. LinkedIn uses these techniques to detect early signals of potential spam content and takes appropriate actions such as filtering or conducting a manual human review. On the other hand, reactive defences employ a combination