An attention enhanced sentence feature network for subtitle extraction and summarization
作者：
Highlights：
• A new sentence feature representation for subtitle summarization was developed.
• Sentence context capturing method by using sentence salience prediction was proposed.
• It is the first study to summarize educational videos using deep learning techniques.
• Sentence scoring metrics is newly proposed for extractive summarization.
• The proposed model is superior to prior works, testing on several well-known datasets.
摘要
•A new sentence feature representation for subtitle summarization was developed.•Sentence context capturing method by using sentence salience prediction was proposed.•It is the first study to summarize educational videos using deep learning techniques.•Sentence scoring metrics is newly proposed for extractive summarization.•The proposed model is superior to prior works, testing on several well-known datasets.
论文关键词：Extractive summarization,Subtitle summarization,Integration CNN-LSTM,Multiple attention mechanism,Educational video
论文评审过程：Received 28 February 2020, Revised 25 November 2020, Accepted 23 March 2021, Available online 31 March 2021, Version of Record 20 April 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.114946
