Human content moderating is a big part of Trust & Safety operations - and also a costly one. This lightning talk will explore how to leverage those reviews to create repositories of known bad or safe content, e.g. images and videos - and proactively remove subsequent uploads of similar or exact copies of that content automatically, resulting in cost savings and safer platforms. Topics covered

- Case-studies on savings and automation rates using indexing

- Crash-course on indexing and hash-matching

- Framework on how to set-up operations and measurement required to populate and maintain content repositories

- Potential pitfalls to take into consideration

Lightning talks are 5-7 minute talks grouped with other 5-7 minute talks across a 50 minute session. There will be time for Q&A after all of the lightning talks in a session.