The meeting involved State Changers discussing the moderation of text and uploaded images within an app platform. One participant has recently launched his application but removed it due to concerns about insufficient content moderation, revealing potential for explicit text or images to be posted and consequently, the app possibly being removed by Apple.
The conversation led to discussions on different methods of moderation, such as manual approvals, allowing everything, or using third party APIs, such as adult content checkers. There was also talk about leveraging computer vision technology to identify inappropriate content. One suggestion was to implement a "moderation hold" mode on content uploads. This concept involves uploading content to a holding area for review before it's officially published. Both approaches to achieving this were discussed: either creating a separate table for all pending content or assigning a status to each post and only showing posts with specific statuses. The State Changer guiding the discussion emphasized that the process of checking moderation doesn't pose the main challenge; the critical concern is developing a staged content model. The schema should adjust with user behavior regarding proper content sharing practices and increase in scale. Tools like Xano were discussed as valuable in this context due to the flexibility in creating specific logics between the database and front-end response resulting in simpler implementation. The meeting also touched on the user experience during moderation by introducing the idea of "Shadowban". It allows users to view their posts (even while they are in moderation) but restricts visibility to others until the content has passed moderation. Towards the end of the discussion, different levels of content checks were specified such as pending, needs review, and rejected. The meeting proposed using automated tasks to move between status levels and maintain an efficient moderation process. Keywords that appeared in the discussion are Xano and Firebase. Other concepts like REST APIs, computer vision, and background task were also part of the discourse.
(Source: Office Hours 12/23 )
