This is a must have feature for researchers in machine learning, computer vision, and physics fields.
This is critical to my use of Papers. Has the implementation not been released yet, because it doesn't appear to work in either the desktop or web apps?
More generally, I would like to be able to import metadata from journal pages, doi references, etc.
At least in arXiv's abstract pages, DOM elements are well annotated by classes.
And regex parsing also seems to work well.
If there is some way to register user-defined mapping(*) from page elements to meta information, I will contribute.
(*) e.g. by JSON
"title": [ "xpath://h1[@class='title']/text()", "regex:/^Title:(.*)$/¥1/" ],
It would be great if you would support importing meta-data from arxiv.org for selected papers. Or importing directly from the html page (example: https://arxiv.org/abs/1811.03600) instead of only from the pdf.
28 people like this idea