A recommendation engine for your applications
summary
A recommendation engine is built for an e‑commerce bookshop (paper and digital) using Prediction.io (Apache) with Spark and Mahout, deployed from a Docker/AMI image. Events are queued and batch-processed; four engines are created to support filtering and are retrain engines nightly. The system serves who bought this also bought and you may like lists, uses a cold‑start fallback to bestsellers, is activated after weeks of collection, runs for over a year and reports positive feedback plus a monthly personalized newsletter.
AI-written summary · about
chapters (21)
- 0:00Introduction to recommendation engines
- 1:43How recommendation engines work
- 2:48Non-personalized recommendations
- 4:56Content-based recommendations
- 7:54Collaborative filtering explained
- 13:44Domain-specific recommendations
- 14:50Persistent vs ephemeral recommendations
- 15:33Explicit and implicit user input
- 16:15Choosing PredictionIO for implementation
- 18:24Event server architecture
- 21:23Sending events to the server
- 23:53Engines and recommendation templates
- 26:29Querying recommendations via API
- 27:34Implementation details and events
- 29:29Displaying recommendations to users
- 30:16Solving the cold start problem
- 30:59Project results and feedback
- 31:20Advanced recommendation approaches
- 33:05Alternative solutions and services
- 34:12DIY recommendation libraries
- 35:35Q&A session
related talks
29mA recommendation engine in your PHP applications
31mCreating recommender systems with Spark, Scala and Prediction.Io
55mBuilding a Recommendation Engine with Machine Learning Techniques
32mRecommender Engines - What are they? Where do they live? What do they feed on?
28mThe Power of Personalized Customer Experiences Through AI Innovations
29mThe Challenges of Building a Personalized Recommendation System in E-Commerce
49mAn Introduction to Recommendation Engines
30mParticle Swarm Classification for Next-Gen Recommendation Engines