- Why build recommendation systems?
- Scope and evolution of recsys
- Prediction and Ranking
- Relevance, novelty, serendipity & diversity
- Paradigms in recommendations: Content-based, Collaborative filtering, Knowledge-based, Hybrid and Ensembles
- Key concepts in recsys:
- Explicit vs. implicit feedback
- User-Item matrix
- Domain signals: location, time, context, social
- Why use deep learning for recsys?
- Primer on deep learning
- Traditional vs deep learning approaches
- Examples and use-cases
- Introduction to the case #1
- Environment setup for hands-on session
- Overview of traditional Colloborative-Filtering for recsys
- Primer on deep learning approaches
- Deep matrix factorisation
- Exercise: Recommending items using Colloborative-Filtering
- Feature extraction using deep learning: Embeddings for Hetrogenous data
- Exercise: Recommending items using similarity measures
- Why learning-to-rank? Prediction vs Ranking
- Rank-learning approaches: pointwise, pairwise and listwise
- Deep learning approach to combine prediction and ranking
- Exercise: Recommending items using Learning-to-Rank
- Combining content-based and collaborative filtering
- Primer on Wide & Deep Learning for Recommender Systems
- Exercise: Recommending items using Hybrid recommender
- Adding temporal component: window and decay-based
- Dynamic and Sequential modelling using RNNs/1D CNNs
- Exercise: Recommending items using RNN recommender
- Deploying the recommendation system models
- Measuring improvements from recommendation system
- Improving the models based on the feedback from production
- Architecture design for recsys: Offline, Nearline and Online
- A/B testing for recommendation systems
- Challenges in recsys:
- Building explanations
- Model debugging
- Scaling-out & up
- Fairness, accountability and trust
- Bias in recsys: training data, UI → Algorithm → UI, private
- When not to use deep learning for recsys
- Recap and next steps, Learning Resources