dgb 0.2

Package for 'Towards better dynamic link prediction' paper at NeurIPS 2022, the code provides download, processing, dataloading and evaluation code for a suite of dynamic graph datasets.

Dynamic Graph Processing

Provides

1. A data-loader to download a suite of dynamic graph datasets.
2. An evaluator for the suit of dynamic graph datasets
3. Tools to benchmark graph models, example code snippets, tutorials, etc. (ongoing...)

---

To install

---

use pip for Python, make sure version python version is 3.6+

>>> pip install dgb

Code examples

---

Code snippet to import the module::

import dgb

Code snippets to download a dataset::

enron = dgb.TemporalDataSets(data_name="enron")
enron_dict = enron.process()
train = enron_dict["train"]
test  = enron_dict["test"]
val   = enron_dict["validation"]

to print all possible datasets::

data_list = dgb.list_all()
for data_name in data_list:
print(data_name)

to download all possible datasets that have not been downloaded yet::

dgb.download_all()

to force redownload all datasets::

dgb.force_download_all()

to skip download prompts and dataset statistics when processing::

dgb.TemporalDataSets(data_name="enron", skip_download_prompt=True, data_set_statistics=False)

License

---

MIT License