Closed OlgaGKononova closed 5 years ago
Yes, sometimes I realize the parser generates words like "objects11", which causes problems. It's better that we remove them. #4
Can you find these numbers for RSC? In the case of the text does a direct reference to the number, ex: "As demonstrated in [2]". Is wise to remove it?
When you say removing reference, you're referring to string such as "(2)" and "[10]"?
Yes
I think they are not helpful for our tasks. And I think it's good to remove them all (any kinds of reference).
Ok, Thanks! We are going to try to remove them
Usually, references are included in special HTML tags. In the case of ECS, these tags are "\<a ... class='xref-bibr' ..>\". While in the case of "As demonstrated in [2]", [2] is not marked up with \ tag. Since there are so few such cases, I will advocate only remove references within "\" for now. Especially in the case of "as demonstrated in [2], XXX can be synthesized with ...", losing such links may not be favorable. If we want to deal with this case, maybe build another ingredient to replace <a ... class='xref-bibr' ..> without \ tags with "PAPER" keyword, and record this link in the JSON output.
Also @tiagobotari @nicolas-mingione It will be useful to document which commits fix the issue. https://help.github.com/articles/closing-issues-using-keywords/
Thank you, Nicolas, This could be an issue. Can you test that and give a return about.
My bad, there actually is a tag... I'm submitting a new version today.
Good, thanks.
I also suggest do not store references at all as paragraphs.
solved 087f7a8386821d07eca112072826468c68f934e6. closing issue.
ECS parsed text still contains reference number in the text. Please, remove them. Same for all other parsers.