liuyeah/K-HTC

K-HTC

Source code for our paper: "Enhancing Hierarchical Text Classification through Knowledge Graph Integration" [1]

Data Preparation

HTC dataset

1. BlurbGenreCollection-EN (BGC) Dataset: https://www.inf.uni-hamburg.de/en/inst/ab/lt/resources/data/blurb-genre-collection.html
2. Web-of-Science (WOS)[2] Dataset: https://data.mendeley.com/datasets/9rw3vkcfy4/2

Knowledge Graph

1. We adopt the advanced knowledge graph named ConceptNet [3].

Data Process

1. Following the strategy proposed by KagNet[4] to recognize the concepts in the documents. Please refer to the original codes: https://github.com/INK-USC/KagNet.
2. After recognizing the concepts, adopt the OpenKE[5] to train concept embedding via TransE.

Model

After we obtain the processed data, we could run the main.py (we give the example on WOS dataset). For the convenience of reproduction, we give the input data format:

processed_data
├─wos_train.json
├─wos_valid.json
├─wos_test.json
├─wos_label_name.json
├─wos_label_relation.json
├─id2label.json
├─entity_embedding.json

1. wos_train/valid/test.json:

{
    'token': [1, 2, 3, ..., 78], # token id
    'concept': [0, 0, 1, ..., 0], #corresponding concept id
    'concept_neighbor': [[1, 2], [3, 4], ...], corresponding concept relation
    'labels': [0, 0, 0, 0, 1, 0, ...], # one-hot
}

2. wos_label_name.json:

{
    'token': [[1, 2, 4], ...], # num_label * list
    'concept': [[0, 1, 0], ...], # num_label * list
    'concept_neighbor': [[[1, 2], [3, 4], ...], ...], # num_label * list
}

3. wos_label_relation.json:

[
    [0, 1], 
    [0, 2],
    ..., # [label id, label id] means these two labels have relations
]

4. id2label.json:

{
    "0": "CS",
    "1": "Medical",
    ...
}

5. entity_embedding.json:

# size: num_concepts * concept_embedding_size
[
    [...]
    ...
]

Citation

[1] Ye Liu, Kai Zhang, et al. Enhancing Hierarchical Text Classification through Knowledge Graph Integration[C]//Findings of the Association for Computational Linguistics: ACL 2023. 2023: 5797-5810.
[2] Kamran Kowsari, Donald E Brown, Mojtaba Heidarysafa, Kiana Jafari Meimandi, Matthew S Gerber, and Laura E Barnes. 2017. Hdltex: Hierarchical deep learning for text classification. In 2017 16th IEEE international conference on machine learning and applications (ICMLA), pages 364–371. IEEE.
[3] Robyn Speer, Joshua Chin, and Catherine Havasi. 2017. Conceptnet 5.5: An open multilingual graph of general knowledge. In Thirty-first AAAI conference on artificial intelligence.
[4] Bill Yuchen Lin, Xinyue Chen, Jamin Chen, and Xiang Ren. 2019. Kagnet: Knowledge-aware graph networks for commonsense reasoning. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 2829–2839.
[5] Xu Han, Shulin Cao, Xin Lv, Yankai Lin, Zhiyuan Liu, Maosong Sun, and Juanzi Li. 2018. Openke: An open toolkit for knowledge embedding. In Proceedings of the 2018 conference on empirical methods in natural language processing: system demonstrations, pages 139–144.