This document provides an overview of a real-time directed content system that uses MongoDB, Hadoop, and MapReduce. It describes:
- The key participants in the system and their roles in generating, analyzing, and operating on data
- An architecture that uses MongoDB for real-time user profiling and content recommendations, Hadoop for periodic analytics on user profiles and content tags, and MapReduce jobs to update the profiles
- How the system works over time to continuously update user profiles based on their interactions with content, rerun analytics daily to update tags and baselines, and make recommendations based on the updated profiles
- How the system supports both real-time and long-term analytics needs through this integrated approach.