sav-vishal / biobert_pretrained_model

biobert-test

This code is for medical text summarization by extractive strategy, using bert with a pre-trained model.

The original code is from BERT-based-Summ. The associated paper for the code is Deep contextualized embeddings for quantifying the informative content in biomedical text summarization

The objective of the paper is to show how contextualized embeddings produced by a deep bidirectional language model can be utilized to quantify the informative content of sentences in biomedical text summarization

(5) (PDF) Deep contextualized embeddings for quantifying the informative content in biomedical text summarization. Available from: https://www.researchgate.net/publication/336272974_Deep_contextualized_embeddings_for_quantifying_the_informative_content_in_biomedical_text_summarization [accessed Mar 05 2021]. ontext-sensitive embeddings for biomedical text summarization

Some modifications to the Summarizer.py script have been done.

Due to compatibility issues it works with python3.6.

Setup Script

The setup.sh script is created to automatically download all the dependencies for the project. It follows these steps:

1. Clone bert repository
2. Download pre-trained biobert_v1.1_pubmed model
3. Install python dependencies in a virtualenv
4. Creates directories INPUT OUTPUT TEMP

    sh setup.sh

Run the Summarizer script

1. Locate the input file at INPUT directory
2. Run the script:

Four parameters must be specified when running the script:

• INPUT_FILE_NAME(-i) is the name of input file already copied to the INPUT directory.
• OUTPUT_FILE_NAME(-o) is the name of output file containing the summary that will be created in the OUTPUT directory.
• COMPRESSION_RATE(-c) specifies the size of summary and takes a value in the range (0, 1).
• NUMBER_OF_CLUSTERS(-k) specifies the number of final clusters in the clustering step.

     python3.6 Summarizer.py -i 14.txt -o 14.txt -c 0.5 -k 4