search

RUCAIBox / RecBole-CDR

This is a library built upon RecBole for cross-domain recommendation algorithms
MIT License
82 stars 12 forks source link

[💡SUG] Support for dual-target or multi-target cross-domain recommendation #42

Closed ShadowTinker closed 1 year ago

ShadowTinker commented 1 year ago

Is your feature request related to a problem? Please describe. Current RecBole-CDR only supports single-target CDR models. Will you add support for multi-target CDR problem mentioned in this survey in a future release? Multi-target CDR is to improve the recommendation accuracy in all domains simultaneously rather than just one target domain.

Describe the solution you'd like New dataset, dataloader and trainer supporting multi-target CDR.

Describe alternatives you've considered I find this part in dataset.py. But I still don't know how to set source_split_flag and change single-target CDR to multi-target CDR https://github.com/RUCAIBox/RecBole-CDR/blob/1758db49e193be4ba70b4c0023bba78af6c2a1f9/recbole_cdr/data/dataset.py#L558-L567

Wicknight commented 1 year ago

@ShadowTinker Hello, thanks for your attention to RecBole-CDR! Thank you for your advice! We will consider providing support for multi-target CDR in the new release. As for the source_split_flag in dataset.py, we can not implement multi-target CDR with this parameter. It only provides the training data of the source domain for training, and the ultimate goal is to improve the performance of the target domain. Maybe you can try changing the source domain dataset structure while getting training, validation, test sets, and changing the evaluation method in the trainer.

ShadowTinker commented 1 year ago

@Wicknight Thanks for your reply and kind advice! I'm looking forward to the new release of RecBole-CDR. I'll try to change the code according to your suggestion.