Hypergraphs with Attention on Reviews for Explainable Recommendation

[theisjendal]

Description

I have added a new model named HypAR, a new example file for HypAR, and made edits to Lightgcn.

For HypAR I've created the recom_hyper file, requirements.txt and requirements_cu116.txt. The example file runs the method on the Cellphone dataset. As this is not part of Cornac, I simply define the methods for creating the dataset directly in the example. The parameters for HypAR are found on the validation set.

For LightGCN I have (see commit 1799bf8):

• Removed normalization for layers, not used for lgcn.
• Fixed sum weight constant to num layers instead of cur layer index.
• Allow lgcn to take blocks.
• Fixed requirement error caused by newer dgl versions.
• Moved edge normalization to graph for easier use. Without these LightGCN wouldn't perform as required.

Todo: Add reference to paper once publicised in readme.

Checklist:

• [ ] I have added tests.
• [x] I have updated the documentation accordingly.
• [x] I have updated README.md (if you are adding a new model).
• [x] I have updated examples/README.md (if you are adding a new example).
• [ ] I have updated datasets/README.md (if you are adding a new dataset).

[lthoang]

LGTM

[tqtg]

@theisjendal @lthoang thanks for creating and reviewing this PR. Since there are many changes introduced, I would suggest to create separate PR that is dedicated to the changes in the LightGCN model first. I also would like to learn the motivation behind those changes. After that, we can revisit this PR for the purpose of adding the HypAR model. How does that sound?

[tqtg]

LGTM