Duplicate detection considers articles with same long DOIs dup, but works correctly using short DOIs

JohnAtl

If two articles are different in every way, but have the same long DOI, they will be marked as duplicate. If I convert the DOIs to short DOIs, even though the short DOIs are the same, they will not be considered duplicates.
E.g. two articles have the long DOI: 10.1016/j.clinph.2014.07.026
When I convert to short DOI, they become: 10/f65djr

When the DOI is long, they appear in Duplicate Items. When short, they do not.

adamsmith

Zotero just kind of ignores shortDOIs because they've mostly fallen into disuse, but the behavior for long DOIs is correct. DOIs are unique identifiers, it should just not be possible for two items to have the same DOI (it is undesirable but possible the other way around), and if they do, something is quite fundamentally broken (how would resolution via doi.org work, e.g.???) and you probably shouldn't use the DOI of the item that's not addressed by doi.org/10.1016/j.clinph.2014.07.026

DWL-SDCA

The only times I see the same doi assigned to more than one (different) item is when a doi is assigned to a series of letters to the editor responding to a published article. When I index those letters separately I make minor descriptive changes to the title of each letter. The exception to this is that long ago a book (as a whole) might be assigned a single DOI and each chapter will "use" the whole book's DOI. Elsevier and National Academies Press were publishers that did this.

If you are seeing the same DOI associated with more than one discrete article something is wrong. You should verify both DOIs as one may be incorrect.

[I can find only one article associated with the DOI, 10.1016/j.clinph.2014.07.026]

adamsmith

Good call, yes, letters, front matter, etc. often only have one DOI, so that'd be a valid case for different items with the same DOI. Zotero has long said they'll eventually allow marking items as non-duplicate, and that'd seem like a good use case for that.