Hierarchical Bidirectional Attention-Based RNN in BioCreative VI Precision Medicine Track, Document Triage Task

Repository containing the winning system of the BioCreative VI Precision Medicine Track, Document Triage Task (2017). The model is a Hierarchical Bidirectional Attention-Based RNN, implemented in Keras.

Embeddings

Put (or link) the embeddings file PubMed-w2v.bin into the embeddings directory, or use Cached Embeddings.

Cached Embeddings

Put (or link) the cached embeddings file _word_vectors.p into the embeddings directory.

Run and get saved model

Run triage.py
Wait for the training to stop. The saved model will be stored at triage/models/experiments
Use triage/models/submissions/submission.py to load the saved model and use it to get the predictions on a new dataset. Set model_name to the name of the desired saved model in triage/models/experiments.

Cite

If you use this code please cite us.

ACM style

Aris Fergadis, Christos Baziotis, Dimitris Pappas, Haris Papageorgiou, and Alexandros Potamianos. 2018. Hierarchical bi-directional attention-based RNNs for supporting document classification on protein–protein interactions affected by genetic mutations. Database 2018, (August 2018). DOI:https://doi.org/10.1093/database/bay076

Bibtex

@article{10.1093/database/bay076,
    author = {Fergadis, Aris and Baziotis, Christos and Pappas, Dimitris and Papageorgiou, Haris and Potamianos, Alexandros},
    title = "{Hierarchical bi-directional attention-based RNNs for supporting document classification on protein–protein interactions affected by genetic mutations}",
    journal = {Database},
    volume = {2018},
    year = {2018},
    month = {08},
    issn = {1758-0463},
    doi = {10.1093/database/bay076},
    url = {https://doi.org/10.1093/database/bay076},
    note = {bay076},
    eprint = {https://academic.oup.com/database/article-pdf/doi/10.1093/database/bay076/27438815/bay076.pdf},
}