Cross-Domain Recommendation via Progressive Structural Alignment

About Our Work

Update: 2023/09/14: We have created a repository for the paper titled Cross-Domain Recommendation via Progressive Structural Alignment, which has been submitted to the IEEE Transactions on Knowledge and Data Engineering (TKDE) journal. In this repository, we offer the original sample datasets, preprocessing scripts, and algorithm files to showcase the reproducibility of our work.

Requirements

• Python == 3.8
• Pytorch == 1.11.0
• DGL == 0.9.1
• gensim == 3.8.3
• nltk == 3.7
• stanfordcorenlp == 3.9.1.1

Data Sets

The structure of the data set should be like

Douban
|_ douban_feature_raw
|  |_ bookreviews_cleaned.txt
|  |_ books_cleaned.txt
|  |_ moviereviews_cleaned.txt
|  |_ movies_cleaned.txt
|  |_ music_cleaned.txt
|  |_ musicreviews_cleaned.txt
|  |_ users_cleaned.txt
|_ douban_feature
|_ douban_movie
|_ douban_book
|_ douban_music
Amazon
|_ ...
|_ ...

Due to file size limitations, we have not uploaded all of the data. The Amazon data can be obtained from this website, while the Lenovo data is commercially licensed and requires you to request access from us.

RUN

# unzip all files into the douban_feature_raw directory
# preprocess could be found in GADTCDR.[https://github.com/FengZhu-Joey/GA-DTCDR]
python main.py # main file

Contact

If you have any questions, please contact me via [email protected].

Contributors ✨

Thanks goes to these wonderful people (emoji key): Thanks to the data preprocessing piplines in GADTCDR https://github.com/FengZhu-Joey/GA-DTCDR

Chuang Zhao 🤔 💻

This project follows the all-contributors specification. Contributions of any kind welcome!