Right, so those fall into the category of random documents that just happen to be PDFs. From the support page linked above:
While it can sometimes extract basic information (title, author) from other documents, you shouldn't expect that — anything can be distributed as a PDF, but that doesn’t mean there’s any standard metadata available for it.
Note that you can and should save from the main webpage whenever possible. In the case of the WIPO report, that'll get you ever-so-slightly more info — it'll prefill the title, URL, and abstract, basically — but even that webpage doesn't have basic metadata on it.
If there's a file that you think should be recognized but isn't, provide a link here.