Advanced AI Prompt for Predictive Sentiment Analysis Framework

Act as a Lead NLP Data Scientist for Social Media. Architect a robust technical workflow to continuously ingest, monitor, and predict macro sentiment shifts for a community of over 50k members spread across [DATA_SOURCES] (e.g., Discord, Reddit, Twitter). Programmatically incorporate the [HISTORICAL_CRISIS_DATA] to train the model to identify the specific early-warning linguistic signs of community churn or impending backlash. Output a comprehensive brief including: 1) A highly specific taxonomy of sentiment tags (e.g., Positive, Neutral, Sarcastic, Urgent, Toxic). 2) A mathematical methodology for calculating a real-time 'Community Health Score'. 3) An automated webhook alert system configured to trigger PagerDuty whenever the rolling average drops below the [TARGET_SENTIMENT_SCORE] on a [REPORTING_FREQUENCY] basis.