News for hackers. Stuff that matters.
Re:A Gentle Introduction to Text Summarization in Machine Le... by ColanR on Wednesday April 17, 2019 @11:28AM (#19683414)
See the subsection titled "Abstraction-based summarization" in the article. Basically, instead of copying important sentences verbatim (and spending all your analysis on choosing those sentences), rewrite entire paragraphs into new and unique sentences. No more plagiarism, because nothing is a direct copy.
Re:A Gentle Introduction to Text Summarization in Machine Le... by heyyyouu on Wednesday April 17, 2019 @12:41PM (#19684135)
That's interesting. One thing about plagiarism is that derivative works are considered plagiarism, even if the direct language is no longer used. https://www.copyright.gov/circs/circ14.pdf A court would have to rule, of course, but I wonder if this would actually get around that. That said, as US caselaw currently stands summaries are completely legal to a certain length -- you can't stop someone from summarizing what exists, it's like sharing a fact -- but you can't create a derivative work. So as long as the resulting output was short enough you shouldn't have to worry anyhow about plagiarism, since summary is covered, but if it's longer, derivative could be an issue. Anyway, thank you -- very interesting!
Re:A Gentle Introduction to Text Summarization in Machine Le... by ColanR on Wednesday April 17, 2019 @01:04PM (#19684353)
> So as long as the resulting output was short enough you shouldn't have to worry anyhow about plagiarism

Ya, that's what I was trying to get at when I mentioned summarizing paragraphs into sentences. I think we're on the same page there. :)

Re:A Gentle Introduction to Text Summarization in Machine Le... by heyyyouu on Wednesday April 17, 2019 @06:14PM (#19686925)
Ah, gotcha! :-)