DRAGON

Pytorch implementation for "Enhancing Dyadic Relations with Homogeneous Graphs for Multimodal Recommendation"-arxiv

Overview of DRAGON

Data

Data could be download from Google Drive: Baby/Sports/Clothing

Run the code

1. Download the data from the data link we provided above, then put the download data to the ./data folder
2. Run generate-u-u-matrix.py on the dataset you want to generate the user co-occurrence graph
3. Enter the src folder and run with python main.py -m DRAGON -d dataset_name

The parameters to reproduce the result in our paper

Datasets	learning rate	reg weight
Baby	0.0001	0.001
Sports	0.0001	0.001
Clothing	0.0001	0.1