A Novel Hybrid Approach for Biomedical Text Summarization Using Pre-Trained Language Models
Keywords:
Natural Language Processing (NLP), BART, T5, Pegasus, Biomedical Text Summarization (BTS)Abstract
The rapid expansion of biomedical literature requires efficient methods for information extraction.This paper proposes a hybrid biomedical text summarization (BTS) system that leverages the strengths of multiple pre-trained language models (PLMs): BART, Pegasus, and T5. Unlike existing approaches that rely on a single PLM, our hybrid system combines the output of these models, employing a novel sentence scoring mechanism and subsequently removing redundant information using cosine similarity to enhance the quality and conciseness of the summaries
References
. Lalitha, E., et al. (2023). Text Summarization of Medical Documents using Abstractive Techniques. 2nd International Conference on Applied Artificial Intelligence and Computing (ICAAIC).
. Lewis, M., et al. (2020). BART: Denoising sequence-to-sequence pre-training for natural language generation, translation, and comprehension. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics.
