I’m making a project where I need an algorithm that summarizes text using own words. I found a few codes which chose the best sentences but that is not what I wanted. So can human still beat the best algorithms and is there papers or github projects I should read? I think everything is changing so fast that even 4 years old articles are already old.
This paper shall give you a good introduction on the topic : http://jad.shahroodut.ac.ir/article_1189_8520fe22d84709b4fcf117b3c60a1c0b.pdf
Amongst products, I am currently reviewing Resoomer and Autosummarizer. Wish to find a solution that addresses similar quality using FastAI RNN and transfer learning some day. The first step would be to see a solution that can perform sentence extraction based single document summarization without factual errors and benchmark them on CNN/DailyMail datasets.
Can you please point out some resources that choose the best sentences?