Download the data from https://drive.google.com/file/d/1os3Kmk6iofy_0Yf0nskglEVwOpKNhPxx/view?usp=sharing and uncompress it under data directory. We encourage you to split the datasets into training and test data yourself and store them like the demo dataset.
First enter model/BSS-GED and run ./setup.sh Then enter main, run make (It is very likely that you will need to change the value of the cflags and ldflags in main/makefile. Their values should be the same as the outputs of command python3-config --cflags and python3-config --ldflags.
First make sure configure config.py correctly. Then run python TrainModel.py
Enter main. Then run bin/main with correct commands.
The command format is: bin/main train_graph_file train_graph_number test_graph_file test_graph_number model_path, index_path (GED2Hamming mapping file)
The commands to run AIDS, linux15, ALCHEMY in our experiment is shown below. Make sure the parameters in main.cpp are consistent with the parameters in config.py.
| Dataset | command |
|---|---|
| AIDS | bin/main ../data/AIDS/train/graphs.bss 42587 ../data/AIDS/test/graphs.bss 100 15 model_path inverted_index_path |
| linux15 | bin/main ../data/linux15/train/graphs.bss 24087 ../data/linux15/test/graphs.bss 100 15 model_path inverted_index_path |
| ALCHEMY | bin/main ../data/ALCHEMY/train/graphs.bss 99676 ../data/ALCHEMY/test/graphs.bss 100 15 model_path inverted_index_path |
You need to make sure the dataset specified in model/config.py is consistent with the dataset in the commands above.
- python3: Assume python3 by default (use pip3 to install packages).
- numpy
- pandas
- scipy
- scikit-learn
- tensorflow
- networkx==1.11 (NOT 2.1)
- matplotlib
Reference commands: sudo pip3 install numpy pandas scipy scikit-learn tensorflow networkx==1.11 matplotlib