These are the datasets used in "MGTCOM: Community Detection in Multimodal Graphs" The dataset preparation code can be found in our repository. Each dataset consists of a heterogenous graph with additional edge or node timestamps and preprocessed feature vectors. For each dataset, the files are split into raw and processed folders. * `raw` folder: contains a set of parquet files with formatted raw dataset data. Files follow the naming convention `node_<name>` or `edge_<from>_<rel_name>_<to>`. * `processed` folder: contains preprocessed datasets in pytorch geometric graph data format