This is the model in "A Graphical and Attentional Framework for Dual-Target Cross-Domain Recommendation" (IJCAI2020). GA-DTCDR is an optimized model for DTCDR ("DTCDR: A Framework for Dual-Target Cross-Domain Recommendation" in CIKM2019). DTCDR is the first work for dual-target cross-domain recommendation. Compared with DTCDR, we improved the embedding strategy (from DMF/NeuMF to Graph Embedding) and combination strategy (from fixed combination operators to element-wise attention).
As for the doc2vec code and the raw data including text information, I have shared the desensitization raw data at https://www.researchgate.net/publication/350793434_Douban_dataset_ratings_item_details_user_profiles_and_reviews. If you want to learn how to use Doc2vec, you can visit https://radimrehurek.com/gensim/models/doc2vec.html#gensim.models.doc2vec.Doc2Vec.
Due to the size limit (the file size of raw dataset is too large), so I upload the raw dataset at ResearchGate.
If you want to use our code or dataset, you should cite the following papers (at least one paper) in your submissions.
@inproceedings{zhugraphical, title={A Graphical and Attentional Framework for Dual-Target Cross-Domain Recommendation}, author={Zhu, Feng and Wang, Yan and Chen, Chaochao and Liu, Guanfeng and Zheng, Xiaolin}, booktitle={Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI 2020}, pages={3001--3008}, year={2020} }
@inproceedings{zhu2019dtcdr, title={DTCDR: A framework for dual-target cross-domain recommendation}, author={Zhu, Feng and Chen, Chaochao and Wang, Yan and Liu, Guanfeng and Zheng, Xiaolin}, booktitle={Proceedings of the 28th ACM International Conference on Information and Knowledge Management}, pages={1533--1542}, year={2019} }
(1) Requirements:
pip install gensim
pip install node2vec
(2) Running:
python GA-DTCDR.py