Thursday, September 04, 2008

When ISSN's disappear, taking DOIs with them

I've been using ISSN's (International Standard Serial Number) to uniquely identify journals, both to generate article identifiers, and as a parameter to send to CrossRef's OpenURL resolver. Recently I've come across journals that change their ISSN, which has fairly catastrophic effects on my lookup tools. For example, the Canadian Journal of Botany has the ISSN 0008-4026, or at least this is what JournalSeek tells me. However, the journal web site tells me that it has been renamed as Botany, with ISSN 1916-2804. The thing is, if I want to look up DOIs for articles published in the Canadian Journal of Botany, I have to use the ISSN for Botany if I want to get a result. Hence, I can't rely on looking up the ISSN for the Canadian Journal of Botany. I've come across this in other journals as well.

WorldCat's xISSN web services provide some tools to help, including a graphical display of the history of a journal and it's ISSN(s). Here is the history for 1916-2790, redrawn using Graphviz. WorldCat use Webdot, which I've written about earlier. If you view the source of the WorldCat page you can get the link to the original dot file.



The problem with these changes is that it makes ISSN's more fragile. Ideally, the original ISSN would be preserved, and/or CrossRef would have a table mapping old ISSN's onto new ones. The rate things are going, I may have to create such a table myself.

4 comments:

Anonymous said...

Re "a table mapping old ISSN's onto new ones": That chart makes it look much more complicated than a simple one-for-one map. Which ISSN do you use if you want to look up a paper in Canad. J. Res.? Or is that still the old pre-split 0366-6581?

Drycafe said...

Funny - this strikes me as looking surprisingly similar to the history graphs for taxonomic names. Except your example doesn't have journals merging again, but maybe there are some even for that.

Roderic Page said...

David, with respect to Canadian Journal of Research, the point is moot as it's not in CrossRef. If it were, the choice of ISSN would depend on what metadata the publisher submitted to CrossRef.

Hilmar, yep, it's much the same problem. I think some publishers have simply lost patience with the complicated history of journal names that they have inherited, and simply decide to treat them as all part of the same series. It's a nice example of the fragility of linking using metadata, or identifiers derived from metadata.

Chris Freeland said...

Welcome to my world, Rod.