Newer PDF-files often have the doi or a html link to the webpage on the front page.
A nice feature would be to make a translator for pdf-files like this:
detectweb:
-uses the pdftotext function on the frontpage of the pdf
-searches (regex) the resulting textfile for a doi or link to webtext page
-loads the webtext page in the background and runs zotero on this page
-if the fulltext page has a translator return a zotero import icon
doweb:
-get the reference from the fulltext page and add the pdf as attachment.

The tools to do this seems to be there in the zotero code, but unfortunately my coding abilities are not good enough to make a hack demonstrating this.