This is a brief summary of paper for me to study and organize it, Neural Summarization by Extracting Sentences and Words (Cheng and Lapata., ACL 2016) I read and studied.

This paper proposed an extractive summarization method based on neuaral networks.

The extractive unit is two types, one is the sentence and the other is word.

For sentence, extractive summarization method select sentences in a document as summary.

Normally, research of it was consider sequence labeling task to choose sentences in a document

Let’s see an example data

First of all, sentence extractor is as following:

Finally, Word extractor,which is the extracted unit change from sentence to word , is as following:

on evaluation, they used ROUGE-1,2 as means of assessing informativeness and the longest common subsequences(ROUGE-L) as a means of assessing fluency.

Reference

• Paper
  • arXiv Version: Neural Summarization by Extracting Sentences and Words (Cheng and Lapata., arXiv 2016)
  • ACL Version: Neural Summarization by Extracting Sentences and Words (Cheng and Lapata., ACL 2016)
• How to use html for alert
  • how to use icon
• For information
  • Towards Automatic Text Summarization: Extractive Methods on medium
  • Automatic Text Summarization with Machine Learning — An overview on medium
  • Deep Learning Models for Automatic Summarization on toward data science