Recommendation Systems

Recommendation Systems

1. Methods

   • Collaborative Filtering.
     • Memory based Approach
       • User Based
       • Item Based
     • Model based Approach
       • Clustering Technique
         • Nearest Neighbour
       • Association Technique
         • Matrix Factorization
       • Probabilistic Model
         • Bayesian Network
       • Neural Networks
         • Outer Product-based Neural Collaborative Filtering
         • Deep Item-based Collaborative Filtering
   • Content-Based
     • Model based Approach
       • Vector Space Model
       • Probabilistic Model
       • Decision Tree
       • Neural Networks
   • Hybrid Filtering
     • Switched
     • Feature Augmentation
     • Feature Combination
     • Cascade
     • Mixed
     • Weighted
     • Meta Level

2. Evaluation Metrics

   • Statistical accuracy metrics
     • Mean Absolute Error
     • Root-Mean Square Error
     • Correlation
   • Decision Support Accuracy Metrics
     • Recall
     • Precision
   • Hit-rate
   • Diversity Based
     • Pairwise similarity
     • Intra-list diversity

3. Learning to Rank

   • Ranking Mechanism
     • Pointwise
     • Listwise
       • Directly optimize metrics
     • Pairwise
       • Bayesian Personalized Ranking
       • Weighted approximate-rank pairwise
   • top-N recomendations

4. Applications

   • Netflix
   • Amazon/Myntra/Flipkart
   • Grocery
   • Youtube