Which Car Fits My Life? - mobile.de’s Approach to Recommendations
summary
mobile.de describes a hybrid recommender that combines content-based and collaborative approaches, implemented as streaming services on Kafka with backend code in Scala/Java and Cassandra. It applies query, city and inferred family/lifestyle boosts, trains models offline, runs online A/B tests to measure engagement and conversion, and addresses cold-start, messy product-metadata feature engineering, and serving-latency constraints in its deployment decisions.
AI-written summary · about
chapters (8)
- 0:00mobile.de intro
- 3:10Recommending unseen products
- 10:37Pilot projects for preferences
- 17:57Content-based vs collaborative
- 23:12Recommendation engine tech
- 32:22Competing with search
- 37:47Future developments
- 41:26Closing
related talks
43m“Which car fits my life?” - mobile.de’s approach to recommendations
41mRecommender Systems: Crafting Effective Personalised Experiences
55mBuilding a Recommendation Engine with Machine Learning Techniques
43mBuilding your first recommender
27mA smooth ride: Online car buying and selling at mobile.de
28mThe Power of Personalized Customer Experiences Through AI Innovations
29mThe Challenges of Building a Personalized Recommendation System in E-Commerce
01h 14mBuilding a Recommendation System with a Hybrid Search From Scratch