I want to determine the compression rate of scientific articles which I'm going to summarize automatically for my current project. Since, the length of articles varies and I also want to summarize the citation sentences of citing articles to target papers, I need to perform the best approach.
Compression ratio = word count in summary / word count of original.
For example, a paper of 1000 words that is summarised into 100 words has undergone a compression ratio of 1:10 or 10%.
https://www.cs.cmu.edu/~jgc/publication/Summarizing_Text_Documents_Sentence_SIGIR_1999.pdf
There is no best ratio. Too much compression gives incoherent summaries. Too little compression gives little benefit in reading time.
If you want to reduce the length and do not mind if the paper has no references, you can make a popular summary simply by leaving out all citations and the reference list.
Summarising quotations and paraphrases is too dangerous to attempt. You have to let them through unchanged.
I do agree with Professor Ian, plus try to reform your article based on your writing experience.
- Badges
- Science topic
- Similar topics
- Papers