This is our TensorFlow implementation for our paper.
The preprocessed datasets are included in the data (e.g. data/gowalla/gowalla_dataset.csv
), where each line contains an user id
and
an session
meaning an items set (splitted by one day in time order).
For the handling of the file (i.e. data/gowalla/gowalla_dataset.csv
), please refer to this paper and the code(https://github.com/uctoronto/SHAN).
Haochao Ying, Fuzhen Zhuang, Fuzheng Zhang, Yanchi Liu, Guandong Xu, Xing Xie, Hui Xiong, and Jian Wu. *[Sequential recommender system based on hierarchical attention networks]. In IJCAI, 2018.