Interpretable Click-through Rate Prediction through Hierarchical Attention

About

This is the Repo for Interpretable Click-through Rate Prediction through Hierarchical Attention. Please find our paper using the following citation

@inproceedings{han2019all,
  title={Interpretable Click-Through Rate Prediction through Hierarchical Attention},
  author={Li, Zeyu and Cheng, Wei and Chen, Yang and Chen, Haifeng and Wang, Wei},
  booktitle={Proceedings of the Thirteenth ACM International Conference on Web Search and Data Mining},
  year={2020},
  organization={ACM}
}

Run

Dependencies

Please check requirements.txt for dependent packages or run

$ pip install -r requirements.txt

Preprocess dataset

Create following structure in the folder

interhat
|-- data
|   |-- raw
|   |   |-- criteoDAC (put unzipped data)
|   |   |-- avazu (put unzipped data)
|   |-- parse
|-- interhat
    |-- ...

Run preprocess

$ python interhat/preprocess.py [dataset] [n_buckets]

Run `InterHAt`

Run run.sh as an example.

Datasets

This section introduces the datasets.

Criteo

$ curl -O http://azuremlsampleexperiments.blob.core.windows.net/criteo/day_{'seq -s ',' 0 23'}.gz

An useful repo: https://github.com/rambler-digital-solutions/criteo-1tb-benchmark#task-and-data

Avazu

Avazu dataset is from Kaggle: https://www.kaggle.com/c/avazu-ctr-prediction

Frappe

Frappe: https://github.com/hexiangnan/neural_factorization_machine/tree/master/data/frappe

zyli93/InterHAt