Implementation of KGDC

It is the implementation we used for our experiments.

It is not the final version. We will clean up our code and make it more easier to use.

Usage

1. Download Wizard of Wikipedia Dataset and ConceptNet 5.5
2. Run Process.py to process the raw data.
3. Run Datasets.py, that will create a cache file (Remember the directory!)
4. Use the implementation of ConvE to get pre-trained weights.
5. Run main.py. If you want to change settings, check settings/train_config.json、settings/eval_config.json and settings/model_config.json
6. More details can be found in our source code. (But it may be hard to read. Sorry for the bad coding style :( )