Deep Learning for Sequential Recommendation: Algorithms, Influential Factors, and Evaluations

Here is the code and datasets to reproduce the experiment result in our paper "Deep Learning for Sequential Recommendation: Algorithms, InfluentialFactors, and Evaluations"

Code

• GRU4Rec is the code to reproduce the result of basic GRU4Rec, sample size, sample alpha, loss function and data augmentation.
• GRU4Rec-with-dwell-time is the code to reproduce the result of Dwell time.
• GRU4Rec-with-knn is the code to reproduce the result of KNN.
• category-and-behavior is the code to reproduce the result of P-GRU, C-GRU and B-GRU, GRU4Rec(behavior) and GRU4Rec(category).
• attention-mechanism is the code to reproduce the result of Attention mechanism.
• user is the code to reproduce the result of user representation(implicit, embedded and recurrent).
• data_preprocess is the preprocessing code for each dataset.

Datasets

• RSC15 http://www.kaggle.com/chadgostopp/recsys-challenge-2015
• RSC19 http://www.recsyschallenge.com/2019/
• LastFM http://mir.dcs.gla.ac.uk/lastfm/

Experiment result

Please go to https://docs.google.com/spreadsheets/d/1Qs5KKugzheDMFH3YLNoQl50Z3hxRPAvTaEYavGgb5sc/edit?usp=sharing to find more experiment results that we didn't report in the paper.

Acknowledgement

• GRU4Rec is the original Theano implementation GRU4Rec.
• GRU4Rec-with-dwell-time is modified based on the original Theano implementation GRU4Rec.
• GRU4Rec-with-knn is the original code for paper "When Recurrent Neural Networks meet the Neighborhood for Session-Based Recommendation" (http://bit.ly/2nfNldD).
• category-and-behavior is modified based on the TensorFlow implementation GRU4Rec.
• attention-mechanism is modified based on the Pytorch implementation NARM.
• user Implicit and Embedded are modified based on the TensorFlow implementation GRU4Rec. Recurrent is the original code for HRNN (https://github.com/mquad/hgru4rec).