Stale Map Data & Road Closure Detection System
Design a global-scale machine learning system to detect and correct stale map information, specifically focusing on real-time road closures and business hour inaccuracies. The system must ingest billions of daily GPS probes and user reports. Detail the multi-stage architecture for anomaly detection and verification, the data pipelines for real-time processing, the feature engineering strategies for spatial-temporal data, and the evaluation framework for balancing precision (map accuracy) against latency (freshness). Address how to handle label scarcity and the integration of a human-in-the-loop verification process.
XGBoostKafkaFlinkSparkRedisTectonS3Z-score
00