Saving full text directly from ResearchGate

sharachchandra

I don't quite understand why, when I go to Researchgate where someone has uploaded the full-text of their article, the Zotero connector still does not see it as full text, and so I still have to click Download, save the pdf on my laptop, and then upload it to zotero.

(I can share a screenshot if necessary). Am I missing some setting somewhere?

adamsmith

we have an updated RG translator under review, hope to have that back working, to the extent possible, shortly.

Difeorleth

Hi @adamsmith , I (Thomas Boyle: https://www.researchgate.net/profile/Thomas_Boyle2) work at ResearchGate and would like to help facilitate this functionality for researchers using Zotero on RG, particularly for logged-in users which seems to be problematic currently.

adamsmith

Thanks @Difeorleth

I think this actually works better now than it ever has before, but yes, oddly, the experience is worse for logged in users in two ways:

1. Zotero tries to get the item type of an item when loading the page. For not-logged-in users, we're able to get this with one of these two CSS selectors:
span.research-detail-header-section__badge or [data-testid="research-meta-type"]. Neither are available for logged in users. We try
b[data-reactid] instead, but I don't think this is a full equivalent. Having one of the elements available for logged in users -- or another element that's stable across user groups -- would be great.

2. To get to the RIS, we use meta[property="rg:id"], which doesn't exist for logged-in users. We instead try to get this via regex from the URL, which I think typically works, but is obviously more fragile. Again, having the meta property available for logged-in users would make this more robust.

Difeorleth

First off, thanks for answering! So nice to get a chance to sort this out together.

We actually provide scholarlyArticle Schema in JSON-LD format for each publication, both logged in and logged out. Until now, this is what we assumed Zotero Connector used. Perhaps this might be a more robust method of getting the metadata for a publication.

Aside from that, saving a copy of the full-text (when available) fails quite often (compared to logged out). How does the translator trigger that?

adamsmith

Great, thanks -- I can look at the JSON-LD. We might have overlooked it or it's newer than the last real update on the importer. For the PDF, we use
var pdfURL = attr(doc, 'meta[property="citation_pdf_url"]', 'content');
i.e. the citation_pdf_url meta tag in the header. Is there a better option? That one is easy to switch.

Difeorleth

Hi! We've updated our logged in pages so that this hook works as well. Thanks for helping us understand how it works, hopefully this should enable far more researchers to use Zotero and RG more effectively from now on