A Hybrid Model for Medical Paper Summarization Based on COVID-19 Open Research Dataset
Gi Ryung Song, Yongbin Wang · 2020
Automatic generation of summarization or key phrase has been applied in a variety of domains, such as scientific papers and news. In response to the COVID-19 pandemic, the white house and some research groups have prepared the COVID-19 article dataset. To struggle against the COVID-19, the automatic summarization or key phrase method can be useful for those wanting a quick overview of what the latest information is saying on pandemic topics. This paper introduce the COVID-19 dataset from Kaggle and propose a novel model which combine a conventional Seq2Seq model with attention mechanism and a classical keywords extraction method. Our motivation is to obtain key information and maintain the result coherence. Experiment results reveal that our model depending on the COVID-19 dataset achieves a considerable improvement over a classical Seq2Seq model with attention mechanism.