search

openjournals / whedon

Command-line utilities to manage JOSS submissions. May one day be a sentient being.
MIT License
25 stars 32 forks source link

Incorrectly identifies doi to a review of a book when reference is a book #76

Open cicirello opened 4 years ago

cicirello commented 4 years ago

For a paper I recently submitted to JOSS, one of my references is to a book that doesn't have a doi, and whedon incorrectly identifies a doi to a review of the book. Most books (all?) don't have dois. I added the isbn for the book instead. In the pdf generated by whedon (after I added the isbn) the isbn is included as a link to the book's listing in worldcat. That behavior makes sense for books. However, whedon still identifies a doi (during a check references command) to a review of the book. A bibtex entry that can recreate the issue is:

Without the ISBN:

@book{Rusell2009, author = {Russell, Stuart and Norvig, Peter}, title = {Artificial Intelligence: A Modern Approach}, year = {2009}, publisher = {Prentice Hall Press}, address = {USA}, edition = {3rd} }

With the ISBN:

@book{Rusell2009, author = {Russell, Stuart and Norvig, Peter}, title = {Artificial Intelligence: A Modern Approach}, year = {2009}, publisher = {Prentice Hall Press}, address = {USA}, edition = {3rd}, isbn = {0136042597} }

The "check references" command to whedon gives the following message (in both cases above): "https://doi.org/10.1016/0004-3702(96)00007-0 may be missing for title: Artificial Intelligence: A Modern Approach". However, if you follow that doi, you'll discover that it is not for the book itself, but a review of the book authored by someone else that appeared in some journal.

I suggest that the "check references" command shouldn't look for a doi if the reference type is book. And instead, perhaps for books, it should require the ISBN field, especially since it already seems to support linking to worldcat book entries when the ISBN field is present.

arfon commented 4 years ago

Thanks @cicirello. I agree this kind of a bug but it may simply be a limitation of the fuzzy search to the Crossref API that we do with Whedon.

I've added a little extra language to the Whedon response to make it clear these are suggestions only in https://github.com/openjournals/whedon-api/commit/aff5b76cef37e2cdbc884df60abe0539c7daf9ff

cicirello commented 4 years ago

@arfon I think the language update should cover cases like that.