nibuhao / hyperbolic-label-emb-for-hmc Goto Github PK

Joint Learning of Hyperbolic Label Embeddings for Hierarchical Multi-label Classification
Soumya Chatterjee*, Ayush Maheshwari*, Ganesh Ramakrishnan and Saketha Nath Jagaralpudi
European Chapter of the Association for Computational Linguistics (EACL) 2021

environment.yml has the depedencies

Download glove.6B.300d.txt from here in GloVe folder

Please refer to HiLAP for the dataset instructions. Put the required dataset files in folders rcv1, yelp or nyt and run data_utils/gen_json_<dataset>.py for preprocessing the data.

Run main.py using the arguments --exp_name
 --flat for Model_flt
 --cascaded_step1 and --cascaded_step2 for Model_cas
 --joint for Model_jnt

Specify the dataset using --dataset

For examples, please refer Synthetic/all_expts.sh.

• HiLAP for data processing and TextCNN model
• Poincare Embeddings for Poincare utils
• bert-base-uncased-vocab.txt is from Hugging Face Tokenizers

@inproceedings{chatterjee-etal-2021-joint,
    title = "Joint Learning of Hyperbolic Label Embeddings for Hierarchical Multi-label Classification",
    author = "Chatterjee, Soumya and Maheshwari, Ayush and Ramakrishnan, Ganesh and Jagaralpudi, Saketha Nath",
    booktitle = "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume",
    year = "2021",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2021.eacl-main.247",
    pages = "2829--2841",
}