--- name: community-content-moderation description: Moderate, quality-score, and manage community-contributed content in forums and knowledge bases. Use when setting up content moderation workflows, automating quality scoring for community posts, managing contributor programs, handling content disputes, gamifying community contributions, or identifying expert contributors. Triggers on phrases like "content moderation", "community moderation", "community contributions", "content quality scoring", "contributor program", "community expert", "forum moderation", "crowdsourced content". --- # Community-Contributed Content Moderation Manage and quality-control user-generated content in community forums, ensuring high-quality peer support while preventing spam and misinformation. ## Workflow ### 1. Content Ingestion and Initial Screening 1. **Content capture from multiple sources**: - Community forum posts and answers - Knowledge base article suggestions/improvements - In-app feedback and tips submitted by users - Blog comments and discussion threads - Social media responses to support queries 2. **Automated initial screening**: - Spam detection: Check against known spam patterns, repetitive content, promotional links - Profanity and inappropriate content filter - Language detection and auto-categorize by language - PII detection: Flag posts containing customer personal data (email, phone, API keys) - Plagiarism check: Compare against existing KB articles and community posts (>85% match = potential duplicate) 3. **Risk scoring (0-100)**: - Score <20: High risk → auto-reject, notify user with guidelines - Score 20-50: Medium risk → flag for manual review - Score 50-80: Standard → route to moderation queue - Score >80: High quality → fast-track to approved queue ### 2. Quality Assessment and Scoring 1. **Content quality dimensions**: - **Completeness (0-25)**: Does the content fully address the question? Are edge cases covered? - **Accuracy (0-25)**: Is the information factually correct? Cross-reference with official documentation - **Clarity (0-25)**: Is it well-organized, readable, free of typos? Uses proper formatting? - **Originality (0-25)**: Does this add new information vs duplicating existing content? - **Helpfulness signal**: Has other community members upvoted or marked as solution? 2. **Automated quality scoring**: - NLP analysis for completeness (keyword coverage vs known solution patterns) - Cross-reference accuracy against official KB articles and product documentation - Readability scoring using Flesch-Kincaid (target: grade 8 or below) - Formatting check: proper headings, code blocks, links, images - Duplicate detection against existing community content and KB articles 3. **Reviewer assignment**: - Simple content (score >70): Auto-approve with "Community Verified" badge - Moderate content (score 40-70): Route to community moderator - Technical content (API, integrations): Route to subject matter expert (SME) - Content with flagged inaccuracies: Route to content specialist for fact-check - SLA: 24 hours for standard, 12 hours for high-traffic forum posts ### 3. Approval, Publishing, and Attribution 1. **Multi-tier approval workflow**: - Tier 1 (auto-approve): Content scoring >80, no flagged issues, contributor has "Trusted" status - Tier 2 (moderator review): Content scoring 50-80, or new contributor - Tier 3 (SME review): Technical content, content touching API/code, or any flagged issue - Tier 4 (content manager review): KB article submissions, high-visibility content 2. **Contribution feedback to user**: - Approved: Notify user, award points/badge, show published content - Approved with edits: Show what was changed, explain improvements - Needs revision: Return with specific feedback (e.g., "Add screenshots for steps 3-5") - Rejected: Explain reason clearly, suggest how to improve, encourage resubmission 3. **Published content management**: - Tag as "Community Contributed" with author attribution - Track views, helpfulness votes, and ticket deflection - Promote high-performing community content to official KB (with author credit) - Auto-update content when product changes detected - Notify original author of performance metrics and promotion ### 4. Contributor Program Management 1. **Tiered contributor levels**: - **Newcomer (0-10 points)**: Basic posting privileges, auto-review all content - **Regular (11-50 points)**: Can edit other posts, reduced review friction - **Expert (51-200 points)**: Fast-track approval, can moderate others, "Expert" badge - **Champion (201+ points)**: Auto-approve simple content, featured contributor profile, swag rewards 2. **Point and reward system**: - Post accepted answer: +10 points - Article suggestion published: +25 points - Content marked "Best Answer": +15 points per occurrence (max 3) - Helpful votes received: +1 point per vote (max +20/day) - Content promoted to KB: +50 points - Monthly top contributor: +100 bonus points + featured spotlight 3. **Contributor analytics and recognition**: - Monthly contributor leaderboard (public) - Quarterly "Community Champion" awards (featured in company newsletter) - Annual contributor conference invitation or VIP event - Exclusive access to product roadmap discussions for top contributors - Direct Slack channel access to product team for Champions ## Templates & Frameworks ### Content Moderation Queue Dashboard ``` COMMUNITY MODERATION DASHBOARD — June 2025 ============================================ QUEUES: Pending Review: 34 (target: <50) Auto-Approved (last 7 days): 87 Flagged for Manual Review: 12 Returned for Revision: 8 Rejected: 3 Approved and Published: 92 MODERATION METRICS: Average review time: 6.2 hours (target: <12 hours) ✓ Approval rate: 72% (vs 68% last month) ✓ Content quality score (avg): 78/100 False positive rate (good content rejected): 2.1% ✓ Spam caught by auto-screening: 156 (98.5% accuracy) CONTENT SOURCES: Forum posts: 42 KB article suggestions: 18 In-app feedback: 12 Blog comments: 5 Total: 77 new submissions this month TOP CONTRIBUTING MEMBERS (June): Rank | Member | Points | Contributions | Avg Quality | Status -----|-----------------|---------|---------------|-------------|-------- 1 | sarah_dev | 1,247 | 23 | 92/100 | Champion 2 | api_guru_mike | 986 | 19 | 89/100 | Expert 3 | support_steve | 743 | 16 | 85/100 | Expert 4 | newuser_jessica | 156 | 8 | 78/100 | Regular 5 | design_dave | 134 | 7 | 82/100 | Regular CONTENT TYPE BREAKDOWN: How-to guides: 32 (42%) Troubleshooting: 24 (31%) API documentation: 11 (14%) Best practices: 6 (8%) Feature requests discussion: 4 (5%) REJECTION REASONS: Incomplete/inaccurate information: 42% Duplicate of existing content: 28% Promotional/spam content: 18% Inappropriate tone/language: 7% Outdated information: 5% AUTO-MODERATION EFFECTIVENESS: Spam auto-caught: 156 / 158 total spam attempts (98.5%) False positive blocks (good content caught): 2 (1.3%) PII auto-detected and redacted: 7 Plagiarism flags raised: 3 (all confirmed duplicates) ``` ### Contributor Tier Benefits Matrix ``` CONTRIBUTOR TIER SYSTEM ======================== NEWCOMER (0-10 points): Posting permissions: Community forum only Review process: All content reviewed by moderator Badges: "Newcomer" Benefits: Access to community guidelines, welcome tutorial Requirements: Complete community onboarding tutorial REGULAR (11-50 points): Posting permissions: Forum + KB suggestions + blog comments Review process: Simple posts fast-tracked (auto-approve if score >80) Badges: "Regular Contributor" + topic-specific badges Benefits: Edit other members' drafts, vote on content Requirements: 3 accepted contributions within 30 days EXPERT (51-200 points): Posting permissions: All platforms + featured articles Review process: 50% fast-tracked, technical content still reviewed Badges: "Community Expert" + topic mastery badges (API, Billing, Setup) Benefits: Moderate newcomer content, featured profile, direct Slack access Requirements: 10 accepted contributions, avg quality score >80 CHAMPION (201+ points): Posting permissions: All + auto-publish simple content Review process: Auto-approve simple posts, SME review for technical Badges: "Community Champion" (gold) + all earned badges Benefits: - Featured in company blog/newsletter - Early access to beta features - Quarterly 1:1 call with product team - Conference/event invitations - Exclusive swag and rewards - "Trusted" status (near-zero review friction) Requirements: 25+ accepted contributions, avg quality >85, 12-month active status REWARDS SCHEDULE: Monthly: Top 3 contributors get digital badge + email recognition Quarterly: Community Champion award ($50 gift card + newsletter feature) Annually: Top contributor receives $500 reward + conference invitation Lifetime: 1,000+ point contributors get lifetime "Hall of Fame" status ``` ### Content Quality Rubric ``` COMMUNITY CONTENT QUALITY RUBRIC (100 points) ============================================== COMPLETENESS (25 points): 20-25: Fully addresses the question, covers edge cases, provides alternatives 15-19: Addresses core question, minor edge cases missing 10-14: Partial answer, leaves follow-up questions unanswered 5-9: Bare minimum answer, significant gaps 0-4: Incomplete or irrelevant ACCURACY (25 points): 20-25: Factually correct, verified against documentation, up-to-date 15-19: Generally correct, minor inaccuracies 10-14: Some incorrect information or outdated steps 5-9: Significant inaccuracies that could mislead users 0-4: Completely incorrect or harmful information CLARITY AND FORMATTING (25 points): 20-25: Professional formatting, clear headings, code blocks, screenshots, easy to scan 15-19: Good formatting, readable, minor style issues 10-14: Basic formatting, somewhat readable, lacks structure 5-9: Poor formatting, hard to follow, wall of text 0-4: Unreadable, no structure ORIGINALITY (25 points): 20-25: Unique insight, new approach, adds value beyond existing content 15-19: Adds new perspective or updates existing content 10-14: Rephrases existing content with minor additions 5-9: Largely duplicates existing content 0-4: Direct copy of existing content (plagiarism) APPROVAL THRESHOLDS: ≥80: Auto-approve (if contributor has Trusted status) or fast-track 60-79: Publish with minor suggested edits 40-59: Return for revision with specific feedback <40: Reject with explanation and improvement suggestions ``` ## Edge Cases - **Misinformation with high votes**: Community upvotes a wrong answer because it seems helpful: - Monitor "accepted answer" vs actual ticket resolution (did this actually solve the issue?) - If subsequent tickets on same topic indicate wrong solution, depromote and flag for review - Send alert to SME for fact-check - Add "Under Review" banner to content while being verified - Track and update: 28% of high-vote incorrect answers were caught this way - **Contributor conflicts and disputes**: Two contributors edit same content with conflicting information: - Lock content from editing during dispute - Route to senior moderator for arbitration - Both contributors receive feedback on accepted version - Document resolution in moderation log - Escalate to content manager if involves repeated conflict - **Commercial content disguised as help**: Users posting competitor recommendations or promotional content: - AI detection for promotional language patterns - Link analysis: flag posts with multiple outbound links - Cross-reference against competitor name database - Auto-reject and ban repeat offenders after 3 violations - Maintain contributor code of conduct with clear enforcement policy - **High-traffic contributor abuse**: Popular contributors with Trusted status begin posting lower quality: - Quarterly quality audit of all Trusted contributors - If average quality drops below 70/100 for 30 days, downgrade tier - Notify contributor with specific improvement areas - Offer mentorship or content writing guidelines refresh - Allow appeal process with re-evaluation - **Multilingual content moderation**: Contributors posting in languages moderators don't speak: - AI-powered translation for initial screening (with confidence score) - Route to bilingual moderators for languages with dedicated coverage - Community-based peer review for languages without dedicated moderators - Use professional translation service for content promoted to official KB - Track moderation coverage gaps by language and recruit bilingual moderators ## Integration Points - **Community platforms** (Discourse, Vanilla Forums, Circle): Source content for moderation; publish approved content; manage contributor profiles - **Knowledge base systems** (Zendesk Guide, Confluence): Promote community content to official KB; sync content updates - **Help desk systems** (Zendesk, Intercom): Cross-reference community solutions with ticket resolutions; track deflection - **Social media platforms**: Monitor brand community groups (Facebook Groups, LinkedIn Communities) for content opportunities - **Analytics platforms** (Google Analytics, Mixpanel): Track community content performance, user engagement, deflection metrics - **AI/NLP services** (OpenAI, Google NLP, DeepL): Spam detection, quality scoring, plagiarism checking, multilingual content analysis - **Gamification platforms** (Badgeville, custom): Manage points, badges, leaderboards, and rewards - **Email/notification systems**: Send contributor feedback, recognition emails, moderation alerts ## Output ### Monthly Community Content Report ``` COMMUNITY CONTENT REPORT — June 2025 ====================================== CONTENT VOLUME: New community submissions: 77 (↑ 12% from May) Approved and published: 55 (71.4% approval rate) Auto-approved: 87 (from backlog, avg quality 86/100) Returned for revision: 8 (10.4%) Rejected: 14 (18.2%) QUALITY METRICS: Average quality score: 78/100 (↑ 3 pts from May) Content with ≥90 score: 18 (25% of approved) Content promoted to official KB: 6 Average helpfulness votes per post: 4.2 Community content deflection rate: 31% (estimated tickets prevented) CONTRIBUTOR PROGRAM: Active contributors (submitted in last 30 days): 47 New contributors (first submission this month): 12 Champions: 3 Experts: 8 Regulars: 28 Newcomers: 47 BUSINESS IMPACT: Community-solved issues (no ticket created): ~890/month Estimated cost savings: $12,905/month ($14.50 avg ticket cost) Top 10 community experts solve 43% of community-submitted solutions Community content pages: 342 total, 1,240 avg monthly views Average time to first community answer: 2.1 hours (vs 45 min agent SLA) RECOMMENDATIONS: 1. Recruit 2 more bilingual moderators (Japanese, German gaps identified) 2. Launch "New Contributor Challenge" to onboard 15 newcomers this quarter 3. Create API documentation contributor guide (high demand, low supply) 4. Review and update spam detection rules (3 missed spam attempts this month) 5. Expand Champion benefits to include annual in-person contributor summit ```