- Matrix Factorization Techniques for Recommender Systems
- MF
- Factorization Machines
- FM
- Probabilistic Matrix Factorization
- PMF
- Collaborative Filtering for Implicit Feedback Datasets
- WMF / IMF
- Collaborative Filtering with Temporal Dynamics
- SVD++, timeSVD++
- Bayesian Probabilistic Matrix Factorization using Markov Chain Monte Carlo
- BPMF
- Restricted Boltzmann Machines for Collaborative Filtering
- RBM-CF
- Learning to Rank with Nonsmooth Cost Functions
- LambdaRank
- SLIM: Sparse Linear Methods for Top-N Recommender Systems
- SLIM
- Real-time Collaborative Filtering Recommender Systems
- RCF / LSH-CF
- Local Low-Rank Matrix Approximation
- LLRMA
- FISM: Factored Item Similarity Models for Top-N Recommender Systems
- FISM
- Using Graded Implicit Feedback for Bayesian Personalized Ranking
- BPR++
- Logistic Matrix Factorization for Implicit Feedback Data
- LogisticMF
- AutoRec: Autoencoders Meet Collaborative Filtering
- AutoRec
- Collaborative Deep Learning for Recommender Systems
- CDL
- PU Learning for Matrix Completion
- ShiftMC
- Relational Stacked Denoising Autoencoder for Tag Recommendation
- RSDAE
- Reversed CF: A fast collaborative filtering algorithm using a k-nearest neighbor graph
- RCF
- Deep Neural Networks for YouTube Recommendations
- Learning to Rank: From Pairwise Approach to Listwise Approach
- Collaborative Denoising Auto-Encoders for Top-N Recommender Systems
- CDAE
- Variational Graph Auto-Encoders
- VGAE
- code
- Collaborative Recurrent Autoencoder: Recommend while Learning to Fill in the Blanks
- CRAE
- A Neural Autoregressive Approach to Collaborative Filtering
- CF-NADE
- Neural Autoregressive Collaborative Filtering for Implicit Feedback
- implicit NADE
- Item2Vec: Neural Item Embedding for Collaborative Filtering
- item2vec
- Session-based Recommendations with Recurrent Neural Networks
- GRU4Rec
- Exploiting Explicit and Implicit Feedback for Personalized Ranking
- MERR_SVD++
- Improving Top-N Recommendation with Heterogeneous Loss
- Neural Collaborative Filtering
- NCF
- Collaborative Variational Autoencoder for Recommender Systems
- CVAE
- Graph Convolutional Matrix Completion
- GC-MC
- Training Deep AutoEncoders for Collaborative Filtering
- DeepRec
- Deep Matrix Factorization Models for Recommender Systems
- DMF
- Hybrid Recommender System based on Autoencoders
- CFN
- Mixture-Rank Matrix Approximation for Collaborative Filtering
- MRMA
- Neural Factorization Machines for Sparse Predictive Analytics
- NFM
- DeepFM: A Factorization-Machine based Neural Network for CTR Prediction
- DeepFM
- Personalizing Session-based Recommendations with Hierarchical Recurrent Neural Networks
- HRNN / HGRU
- Attentive Collaborative Filtering: Multimedia Recommendation with Item- and Component-Level Attention
- ACF
- Variational Autoencoders for Collaborative Filtering
- Mult-VAE, Mult-DAE
- GraphRNN: Generating Realistic Graphs with Deep Auto-regressive Models
- GraphRNN
- Graph Convolutional Neural Networks for Web-Scale Recommender Systems
- PinSAGE
- Self-Attentive Sequential Recommendation
- SASRec
- Collaborative Filtering with User-Item Co-Autoregressive Models
- CF-UIcA
- Efficient K-NN for Playlist Continuation
- Top-K Off-Policy Correction for a REINFORCE Recommender System
- Buy It Again: Modeling Repeat Purchase Recommendations
- Embarrassingly Shallow Autoencoders for Sparse Data
- EASE
- Knowledge Graph Convolutional Networks for Recommender Systems
- KGCN
- RecVAE: a New Variational Autoencoder for Top-N Recommendations with Implicit Feedback
- RecVAE
- Enhancing VAEs for Collaborative Filtering: Flexible Priors & Gating Mechanisms
- H+Vamp Gated
- RecWalk: Nearly Uncoupled Random Walks for Top-N Recommendation
- RecWalk
- Semi-supervised Learning with Graph Learning-Convolutional Networks
- GLCN
- Session-based Social Recommendation via Dynamic Graph Attention Networks
- DGRec
- Hierarchical Temporal Convolutional Networks for Dynamic Recommender Systems
- HierTCN
- KGAT: Knowledge Graph Attention Network for Recommendation
- KGAT
- code
- Collaborative Similarity Embedding for Recommender Systems
- CSE
- BERT4Rec: Sequential Recommendation with Bidirectional Encoder Representations from Transformer
- BERT4Rec
- Reinforcement Learning to Optimize Long-term User Engagement in Recommender Systems
- FeedRec
- N2VSCDNNR: A Local Recommender System Based on Node2vec and Rich Information Network
- N2VSCDNNR
- Sampling-Bias-Corrected Neural Modeling for Large Corpus Item Recommendations
- A Context-Aware Citation Recommendation Model with BERT and Graph Convolutional Networks
- Cold-start Playlist Recommendation with Multitask Learning
- Music Playlist Recommender System
- RACT: Towards Amortized Ranking-Critical Training for Collaborative Filtering
- RACT
- $S^3$-Rec: Self-Supervised Learning for Sequential Recommendation with Mutual Information Maximization
- S3-Rec
- Efficient Neural Matrix Factorization without Sampling for Recommendation
- ENMF
- code
- Inductive Matrix Completion Based on Graph Neural Networks
- IGMC
- Deoscillated Graph Collaborative Filtering
- DGCF
- Sampler Design for Implicit Feedback Data by Noisy-label Robust Learning
- BPO, NBPO
- Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation
- MetaHIN
- TRSM-RS: A Movie Recommender System Based on Users' Gender and New Weighted Similarity Measure
- TRSM-RS
- Neural Collaborative Filtering vs. Matrix Factorization Revisited
- A Hybrid Approach to Enhance Pure Collaborative Filtering based on Content Feature Relationship
- Content-Based Personalized Recommender System Using Entity Embeddings
- On Mitigating Popularity Bias in Recommendations via Variational Autoencoders
- Bilateral Variational Autoencoder for Collaborative Filtering
- BiVAE
- Recommender Systems with Random Walks: A Survey
- Use of Deep Learning in Modern Recommendation System: A Summary of Recent Works
- On the Difficulty of Evaluating Baselines
- Deep Learning based Recommender System: A Survey and New Perspectives
- Graph Neural Networks in Recommender Systems: A Survey
- Deep Learning on Knowledge Graph for Recommender System: A Survey
- A Survey of Similarity Measures for Collaborative Filtering-Based Recommender System
- A survey on Adversarial Recommender Systems: from Attack/Defense strategies to Generative Adversarial Networks
- Attention Is All You Need
- Transformer
- Semi-Supervised Classification with Graph Convolutional Network
- GCN
- graph2vec: Learning Distributed Representations of Graphs
- Graph2vec
- Poincaré Embeddings for Learning Hierarchical Representations
- Poincaré Distance
- Inductive Representation Learning on Large Graphs
- GraphSAGE
- FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling
- FastGCN
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- BERT
- ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
- ALBERT
- Network of Evolvable Neural Units: Evolving to Learn at a Synaptic Level
- ENU
- Adaptive Estimation for Approximate k-Nearest-Neighbor Computations
- How Powerful are Graph Neural Networks?
- Stochastic bandits with vector losses: Minimizing $\ell^{\infty}$-norm of relative losses
- Accelerating Large-Scale Inference with Anisotropic Vector Quantization
- ScaNN
- MovieLens data 1M/10M/20M/25M
- Study the performance of models under the strictly strong generalization
- Split all users disjointly into train/valid/test set
- For each user in valid/test set, split user feedbacks chronologically into query/relevant set
- Build top-N recommendation systems and evaluate them with various metrics using Google Colab
- pandas
- numpy
- scikit-learn
- scipy
- networkx
- tensorflow 2.4
| Model | Comment |
|---|---|
| ItemPop | Base model, the worst diversity |
| MBCF | Base model of user-based CFs |
| EASE | The best performance, improved model of SLIM |
| AutoRec | Base model of AE |
| DeepRec | Capacity improved model of AutoRec |
| CDAE | Corrupt inputs for robust AE model |
| Mult-VAE | Generative version of AE model with multinomial assumption |
| Mult-DAE | Corrupt input data for robustness with multinomial assumption |
| NCF | A neural extension of MF |
| Item2Vec | Extract item representations |
| kNN | Search k-nearest users |
| RBM | Grandma of AE |
| NADE | AE-based model using ordinal information |
| GRU4Rec | Session based model with GRU |
| HierTCN | Session based model with GRU & 1D-CNN |
| Node2Vec | Item2Vec with Random Walk |
| GCN | Extract graph representation inductively |
| RankSVM | Support vector machine for learning to rank |
| Model | Recall@10 | Precision@10 | HR@10 | nDCG@10 |
|---|---|---|---|---|
| ItemPop | 0.077 | 0.054 | 0.327 | 0.071 |
| SIM | 0.095 | 0.065 | 0.376 | 0.088 |
| *EASE | 0.137 | 0.094 | 0.520 | 0.123 |
| AutoRec | 0.129 | 0.075 | 0.528 | 0.119 |
| DeepRec | 0.105 | 0.075 | 0.476 | 0.097 |
| Mult-VAE | 0.119 | 0.083 | 0.466 | 0.108 |
| RBM | 0.120 | 0.086 | 0.517 | 0.111 |
| NADE | 0.119 | 0.083 | 0.447 | 0.090 |
| RankSVM | 0.107 | 0.076 | 0.432 | 0.096 |