This is our implementation for the paper:
Shaoyun Shi, Weizhi Ma, Zhen Wang, Min Zhang, Kun Fang, Jingfang Xu, Yiqun Liu, and Shaoping Ma. 2021. WG4Rec: Modeling Textual Content with Word Graph for News Recommendation. In Proceedings of the 30th ACM International Conference on Information and Knowledge Management (CIKM ’21).
For inquiries contact Shaoyun Shi (shisy13 AT gmail.com).
@inproceedings{DBLP:conf/cikm/ShiMWZFXLM21,
author = {Shaoyun Shi and
Weizhi Ma and
Zhen Wang and
Min Zhang and
Kun Fang and
Jingfang Xu and
Yiqun Liu and
Shaoping Ma},
editor = {Gianluca Demartini and
Guido Zuccon and
J. Shane Culpepper and
Zi Huang and
Hanghang Tong},
title = {WG4Rec: Modeling Textual Content with Word Graph for News Recommendation},
booktitle = {{CIKM} '21: The 30th {ACM} International Conference on Information
and Knowledge Management, Virtual Event, Queensland, Australia, November
1 - 5, 2021},
pages = {1651--1660},
publisher = {{ACM}},
year = {2021},
url = {https://doi.org/10.1145/3459637.3482401},
doi = {10.1145/3459637.3482401},
timestamp = {Mon, 08 Nov 2021 14:07:22 +0100},
biburl = {https://dblp.org/rec/conf/cikm/ShiMWZFXLM21.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}
requirements.txt - The codes can be successfully run with following packages in an Anaconda environment:
pandas==1.3.4
tqdm==4.62.3
pytorch==1.10.0
numpy==1.21.2
scikit-learn==0.23.2
cudatoolkit==10.2.89
python==3.9.7
Other settings with pytorch>=1.3.1 may also work.
The processed datasets can be downloaded from Tsinghua Cloud, in which *.word_graph.pk is the generated word graph. (The codes for word graph construction are in ./src/preprocess for reference).
You should place the datasets in the ./dataset/. The tree structure of directories should look like:
.
├── dataset
│ ├── sogou
│ └── adressa-1w
└── src
├── data_readers
├── data_processors
├── modules
├── models
├── runners
└── utils
- Some running commands can be found in
./command/command.py - For example:
# WG4Rec on Sogou dataset
> cd WG4Rec/src/
> python main.py --model_name WG4RecSogou --dataset sogou --metrics ndcg@5,ndcg@10,hit@5,hit@10 --test_sample_n 99 --lr 0.001 --l2 1e-4 --sent_max 10 --sample_pop 100 --regenerate 0 --gpu 0