CO2Vec: Embeddings of Co-Ordered Networks Based on Mutual Reinforcement

Official implementation of our paper "CO2Vec: Embeddings of Co-Ordered Networks Based on Mutual Reinforcement" (DSAA 2020). CO2Vec is an order representation learning model for co-ordered netwroks.

Dependencies

The core learning model is built using PyTorch

• Python 3.6.3
• PyTorch 0.3.0

Run

To reproduce the results on UNIV dataset, the hyperparameters are set in example.sh.

  bash example.sh

Data

There should be five data files ready in the 'datasets' folder, e.g. datasets/name/

• <name>_split_train.pkl list of training instance in pickle format, each instance is a three tuple for type-A entities: (ent_i, ent_j, label), label is either -1 or 1
• <name>_split_train_e2.pkl list of training instance in pickle format, each instance is a three tuple for type-B entities: (ent_i, ent_j, label), label is either -1 or 1
• <name>_split_train_pos.cross.pkl list of training instance in pickle format, each instance is a four tuple for cross-entity relations from type-A to type-B entities: (ent_i, ent_j, weight)
• <name>_split_train_pos.double.pkl list of training instance in pickle format, each instance is a four tuple for cross-entity relations from type-B to type-A entities: (ent_i, ent_j, weight)

Cite

Please consider cite our paper if you find the paper and the code useful.

@inproceedings{CO2Vec2020,
 author = {Meng-Fen Chiang and
            Ee-Peng Lim and 
            Wang-Chien Lee and                
            Philips Kokoh Prasetyo},
 title = {CO2Vec: Embeddings of Co-Ordered Networks Based on Mutual Reinforcement},
 booktitle = {IEEE International Conference on Data Science and Advanced Analytics (DSAA)},
 year = {2020}
} 

Feel free to send email to [email protected] if you have any questions. This code is modified from ANR.