Content Moderation Workflow Procedure (TS-PROC-001)
1. Purpose
The purpose of this procedure is to describe the systematic workflow for content moderation activities including automated detection, human review, enforcement actions, and quality assurance to ensure consistent and fair application of community guidelines and legal requirements.
2. Scope
This procedure applies to all user-generated content moderation activities on the video streaming platform including videos, comments, live streams, and user profiles. It covers both automated and human moderation processes across all content categories and geographic regions.
3. Overview
This procedure ensures systematic content review through automated pre-screening, risk-based human review, appropriate enforcement actions, and quality assurance validation. The process prioritizes user safety while maintaining fair and consistent moderation decisions across all content types.
4. Procedure
| Step | Who | What |
|---|---|---|
| 1 | Automated Systems | Scan all uploaded content using AI models for prohibited content, harmful material, copyright violations, and age-appropriateness classifications. |
| 2 | Automated Systems | Generate confidence scores and risk assessments for detected policy violations, flagging high-confidence violations and edge cases for human review. |
| 3 | Triage Specialist | Review automated flags and prioritize content for human moderation based on severity, user vulnerability, and content reach potential. |
| 4 | Content Moderator | Conduct detailed human review of flagged content against community guidelines, considering context, cultural factors, and potential harm. |
| 5 | Content Moderator | Make moderation decision (approve, remove, restrict, require age verification) and document reasoning and policy basis for decision. |
| 6 | Senior Moderator | Review complex or high-impact moderation decisions for consistency and accuracy before final enforcement action implementation. |
| 7 | Enforcement Team | Execute moderation decisions including content removal, account actions, user notifications, and appeal rights communication. |
| 8 | User Communication | Send detailed statement of reasons to affected users within 24 hours, explicitly including: (1) the specific policy violation and grounds for the decision, (2) the territorial scope of any restriction (e.g., removed in specific EU countries or globally), (3) whether the decision was made using automated content moderation tools, and (4) clear information about the user’s right to appeal through the internal appeals process, out-of-court dispute settlement options, and judicial redress possibilities as required by DSA Article 17. |
| 9 | Quality Assurance | Conduct random sampling of moderation decisions for accuracy assessment and calibration feedback to moderators. |
| 10 | Data Analytics | Track moderation metrics including accuracy rates, bias indicators, appeal outcomes, and policy effectiveness for continuous improvement. |
| 11 | Appeals Team | Process user appeals through independent review, provide responses within 7 days, and implement reversals when appropriate. |
| 12 | Policy Team | Review moderation trends and outcomes to identify policy gaps, update community guidelines, and improve moderation effectiveness. |
5. Standards Compliance
| Procedure Step(s) | Standard/Framework | Control Reference |
|---|---|---|
| 1-2 | EU Digital Services Act | Art. 16 |
| 1-2 | PCI DSS v4.0 | Req. 12.1 |
| 4-6 | ISO/IEC 27001:2022 | A.7.2.2 |
| 4-6 | PCI DSS v4.0 | Req. 7.1, 8.1 |
| 8 | EU Digital Services Act | Art. 17 |
| 8 | PCI DSS v4.0 | Req. 12.10.1 |
| 9-10 | EU Digital Services Act | Art. 24 |
| 9-10 | PCI DSS v4.0 | Req. 10.6, 12.2 |
| 11 | EU Digital Services Act | Art. 20 |
| 11 | PCI DSS v4.0 | Req. 7.1.1 |
6. Artifact(s)
A comprehensive moderation decision record containing content assessment, policy analysis, enforcement action, user notification, quality review results, and appeal outcome stored in the content moderation system with appropriate audit trails and privacy protections.
7. Definitions
Triage Specialist: Trained moderator responsible for prioritizing flagged content based on risk and impact assessment.
Confidence Score: Numerical assessment of AI model certainty in detecting policy violations or harmful content.
Edge Cases: Content that falls in gray areas between policy compliance and violation requiring human judgment.
Senior Moderator: Experienced content moderator responsible for reviewing complex decisions and ensuring consistency.
Quality Assurance Sampling: Random selection of moderation decisions for accuracy assessment and calibration purposes.
Independent Review: Appeals process conducted by moderators not involved in the original decision.
8. Responsibilities
| Role | Responsibility |
|---|---|
| Content Moderators | Conduct thorough content review, make consistent policy-based decisions, and document reasoning clearly for all moderation actions. |
| Triage Specialists | Prioritize content review queues, assess risk levels, and ensure efficient allocation of human moderation resources. |
| Senior Moderators | Review complex moderation decisions, provide guidance to junior moderators, and ensure consistency across the moderation team. |
| Quality Assurance Team | Monitor moderation accuracy, provide feedback and calibration, and identify training needs and policy improvement opportunities. |
| Appeals Team | Process user appeals independently and fairly, communicate decisions clearly, and identify systemic moderation issues. |
| AI/ML Engineering | Maintain and improve automated detection systems, monitor model performance, and implement bias mitigation measures. |