DOI lookup finds some, but not all, DOIs

I’m using JabRef 5.9 on a Kubuntu machine (very recently upgraded from the old version in the repositories). I have a large citation library, all in one file. Until recently, none of the entries had DOIs attached. I used the Search document identifier online → DOI tool to try to add DOIs. It managed to find and add DOIs to about half the entries. Some of the ones it didn’t find DIOs for are totally understandable (super old, not online, etc). But a lot of them are very recently published, definitely have DOIs, etc. There doesn’t seem to be any pattern to which ones it found DOIs for and which it didn’t. Any ideas of why it might be failing to find some and how I might make it do so? Thanks!

JabRef queries from DOI.org, which then itself queries from https://www.crossref.org/. You could try to check if crossref has stored metadata for those specific DOIs that you fail to find. If Crossref hasn’t, it is no wonder you fail to find the DOI.

Another explanation could be: While is is easy to fetch data which is connected to a unique ID like the DOI, it is much harder to associate a unique ID with arbitrary bibliographic metadata, as a search may find multiple overlapping and competing results and from these determines a “most likely” candidate. It can be that this most likely candidate is an entry without DOI and unfortunately may not necessarily be the “real” one you are looking for.

Hypothetical example:

You have entry B in your local library and try to find the DOI

  • entry B: author x, title xzx, year x, publisher x, doi: to look for

The web search finds only entry A or finds both entry A and B in some kind of online database:

  • entry A: author x, title xyx, year x, publisher x, doi: no
  • entry B: author x, title xzx, year x, publisher x, doi: yes

Result: The online search either only finds entry A or finds A and B, but determines that entry A is the “most likely candidate”, therefore does not deliver the DOI.

Thanks, that’s helpful—in understanding the issue, even if not solving it (I did wonder if it would be basically unsolvable).

I think the second explanation is the right one. There’s a couple of hundred entries that are almost certainly listed at CrossRef (I checked a few arbitrary ones and they are), so I guess the search is struggling to link the bibliographic data to the right DOI. Interestingly, I just ran the lookup again on those entries, and it did actually find a few that it failed to find yesterday. Anyway, I’m a lot closer to having a fully DOI’d library now, and I can make sure to only import with full DOI info in the future. Thanks!