AHAG

This is our implementation for the paper:

Adaptive Hierarchical Attention-Enhanced Gated Network Integrating Reviews for Item Recommendation

AHAG: This is the state-of-the-art method that uti-lizes deep learning technology to jointly model user and item from reviews for item recommendation.

##Cite

If you use the code, please kindly cite the following paper:

@article{liu2020adaptive, title={Adaptive Hierarchical Attention-Enhanced Gated Network Integrating Reviews for Item Recommendation}, author={Liu, Donghua and Wu, Jia and Li, Jing and Du, Bo and Chang, Jun and Li, Xuefei}, journal={IEEE Transactions on Knowledge and Data Engineering}, year={2020}, publisher={IEEE} }

Environments

• python 3.5
• Tensorflow (version: 1.9.0)
• numpy
• pandas

Dataset

In our experiments, we use the datasets from Amazon 5-core(http://jmcauley.ucsd.edu/data/amazon) Pretrained GloVe embeddings obtained from Wikipedia 2014 + Gigaword 5 with 6B tokens used for words.

Example to run the codes

Data preprocessing:

The implemention of data preprocessing is modified based on this

Train and evaluate the model:

python train.py

Misc

The implemention of CNN is modified based on this The implemention of self-attention is modified based on this