Twitter - Spam Violation - Machine Learning Workflow

A mountain of tweets are created every day on Twitter. The team at Twitter is unable to go through each Tweet by hand to ensure it doesn't violate any of its guidelines. Instead machine learning projects are setup that have humans audit a small set of Twitter data that helps teach their AI to recognize guideline breaches. This Twitter workbench allows human auditors to view small batches of data (column 1), view the Twitter home page of the owner of the current Tweet to gather historic Tweet context (column 2) and in column 3 the auditor answers a few questions to help teach the AI.