search

twchen / lessr

Handling Information Loss of Graph Neural Networks for Session-based Recommendation
MIT License
67 stars 19 forks source link

Baseline code #6

Open ddddyla opened 2 years ago

ddddyla commented 2 years ago

Hi. Could you share the baseline code that on your datasets, for example, SR-GNN? Because I cannot run the baseline code on your datasets.😭

twchen commented 2 years ago

For the code of baselines, please refer to the official implementations. To run the baseline code on our datasets, you can convert our datasets to their format. Besically, you just need to augment the sessions and store them as pickle files. You may also need to shift the item IDs because our item IDs start from 0 and theirs may start from 1.

ddddyla commented 2 years ago

For the code of baselines, please refer to the official implementations. To run the baseline code on our datasets, you can convert our datasets to their format. Besically, you just need to augment the sessions and store them as pickle files. You may also need to shift the item IDs because our item IDs start from 0 and theirs may start from 1.

Yes, I try to convert your datasets to their format, such as augment the sessions, shift item IDs and store them as pickle files. But the SR-GNN results I ran on your transformed dataset are quite different from those shown in your paper. So could you share your transformed datasets?