Multi-relational Stacking Ensemble Recommender System using Cinematic Experience
- (2022.01.20) Multi-Relational Stacking Ensemble Recommender System Using Cinematic Experience in IEEE BigComp2022
- Reference code: https://github.com/zhoujf620/Motif-based-inductive-GNN-training
- Author's Paper link: https://arxiv.org/pdf/1706.02263v2.pdf
- Reference code: https://github.com/tobiasweede/rs-via-gnn/tree/main/02_gcmc/movielens
Dependencies
- PyTorch 1.2+
- DGL 0.5 (nightly version)
Data
Movie data
- Rotten Tomatoes data : https://www.kaggle.com/stefanoleone992/rotten-tomatoes-movies-and-critic-reviews-dataset
Training set for BERT
- Sentiment Analysis : https://www.kaggle.com/c/sentiment-analysis-on-movie-reviews
- Emotion Analysis : https://www.kaggle.com/praveengovi/emotions-dataset-for-nlp
How to run
- grid_search_final.ipynb
Results
| Dataset | Our code best of epochs |
|---|---|
| Rotten Tomatoes | 0.8070 |