1
System Metrics
| Metric |
Value |
Status |
| Active Clusters |
-- |
Operational |
| Emerging Topics |
-- |
Monitored |
| Total Keywords |
-- |
Indexed |
| Recent Articles |
-- |
Processed |
| Unread Alerts |
-- |
Pending |
2
Topic Clusters
Semantic groupings of related cryptocurrency topics identified through vector similarity analysis and momentum detection.
3
Trending Keywords
Most frequently mentioned terms across analyzed content sources, ranked by occurrence frequency.
1
Top Gainers (24h)
Cryptocurrencies with highest percentage gains in the last 24 hours from CoinMarketCap.
2
Top Losers (24h)
Cryptocurrencies with largest percentage losses in the last 24 hours from CoinMarketCap.
1
Topic Alerts
Real-time alerts for emerging trends and high-momentum topics detected by the analysis engine.
1
AI-Generated Content Summaries
Automatic summaries and entity extractions generated by Llama-2 AI model from ingested crypto news content.
1
Pipeline Architecture
Four-stage data processing pipeline utilizing Cloudflare Workers, D1 Database, and AI services for real-time crypto trend detection.
Component: Cloudflare Worker (ingestion-worker)
Trigger: Cloudflare Cron (0 */3 * * * - Every 3 hours)
Sources:
• CoinMarketCap API - Top 100 crypto by market cap
• CoinDesk RSS Feed - Latest crypto news articles
• NewsAPI - Crypto keyword search across 50+ sources
Processing: Deduplication, keyword extraction, metadata enrichment
Database: content_items, content_keywords tables (D1)
Queue: processing-queue (Cloudflare Queue)
Status: --
↓
Component: Cloudflare Worker (processor-worker)
Trigger: Queue consumer (processing-queue)
AI Model: @cf/meta/llama-2-7b-chat-int8 (Workers AI)
Capabilities:
• Text summarization (100 tokens, temp 0.7)
• Named entity recognition (cryptocurrencies, projects, people)
• Keyword importance scoring
Rate Limit: 10 requests/minute per worker
Database: ai_summaries, entity_mentions tables (D1)
Queue: analysis-queue (Cloudflare Queue)
Status: --
↓
Component: Cloudflare Worker (analyzer-worker)
Trigger: Queue consumer (analysis-queue)
Embedding Model: @cf/baai/bge-base-en-v1.5 (768-dim vectors)
Clustering:
• Algorithm: Cosine similarity-based hierarchical clustering
• Similarity threshold: 0.65 (configurable)
• Min cluster size: 3 entities
• Max cluster age: 30 days
Momentum Calculation:
• Hourly mention tracking per entity
• Velocity: (mentions_1h / mentions_6h) ratio
• Momentum score: velocity × uniqueness × freshness
Database: topic_clusters, cluster_members, cluster_history, momentum_metrics tables
Status: --
↓
Component: Cloudflare Worker (analyzer-worker)
Detection Criteria:
• Momentum score ≥ 2.0x (configurable)
• Minimum 5 mentions in 24h window
• At least 2 unique sources
• Cluster age < 7 days (for "emerging" status)
Alert Types:
• 🔥 Hot Topics (momentum ≥ 5.0x)
• 📈 Rising Topics (momentum 2.0x - 4.9x)
• ⚡ Flash Spikes (sudden velocity increase)
Alert Rules:
• Cooldown: 3 hours between alerts for same cluster
• Auto-expire: 24 hours if not acknowledged
Database: topic_alerts table, emerging_topics view
Status: --
2
Detailed System Metrics
1
AI Prompt Configuration
Manage and configure AI prompts for content analysis, summarization, and entity extraction. Supports versioning and testing.
2
Test AI Prompt
Test any prompt configuration with the AI model before deploying.
3
Pipeline Data Sources
Configured data sources and ingestion parameters for the trend detection pipeline.
Endpoint: Top 100 cryptocurrencies by market cap
Frequency: Every 3 hours
Data Points: Price, 24h change, market cap, volume
Status: ACTIVE
URL: https://www.coindesk.com/arc/outboundfeeds/rss/
Frequency: Every 3 hours
Parsing: Title, description, publication date
Status: ACTIVE
Query: Cryptocurrency news from top sources
Frequency: Every 3 hours
Language: English
Status: ACTIVE
4
System Configuration
Core system parameters and thresholds for topic detection and clustering.
| Parameter |
Value |
Description |
| Clustering Threshold |
0.65 |
Cosine similarity threshold for topic clustering |
| Momentum Threshold |
2.0x |
Minimum momentum score for emerging topics |
| Velocity Window |
6h |
Time window for velocity calculation |
| Alert Cooldown |
3h |
Minimum time between alerts for same cluster |
| Entity Confidence |
0.7 |
Minimum confidence for entity extraction |