PublicationYear

[Barbara Hirschmann]

Dear Metadata Working Group

I wanted to voice a concern that some of my clients have expressed on the property PublicationYear.
Those clients are institutions that register DOIs for digitised collections, mostly library material like images, autographs etc.
When looking at their metadata we discussed several options for supplying PublicationYear (and Date). I suggested that PublicationYear could, apart from the original publication date of the image, document etc., be the online publication date of the digital object. However, none of them were happy with this solution (considering the importance of PublicationYear for citation purposes) and they all prefered using the original year.
Now with those kind of objects it happens very often that the publication year is unknown or can only be narrowed down to a date range.
Since we had to choose one date, we now always supply the earliest year of a date range as PublicationYear.
However, this solution doesn't really please our clients, because they feel that a citation that only shows the earlier date of a date range is incorrect.

Has this topic come up in your WG and what is your opinion about it?

Best regards, Barbara
ETH Zurich

[Joan Starr]

Hi Barbara,

Thank you this great question. I have two thoughts on this matter. First, can your clients consider this date in terms of this concept "date when the (object) was first made public (available)"? Second, I'll let you know that the Metadata Working Group is currently considering the option of supplying a standard value to use when this (or any other required property) is truly unknown. If you would like to see an example of what I am talking about, look at how the Dublin Kernel handles this: http://dublincore.org/groups/kernel/spec/#anchor16 

Thanks again for this good question.

Joan Starr

Chair, DataCite Metadata Working Group

[Barbara Hirschmann]

Hi Joan

thank you for answering so quickly.
I would like to give an example: http://dx.doi.org/10.7891/e-manuscripta-679
The creation date of this manuscript is not known exactly, but can be narrowed down to 1490-1499.
Following your definition of PY we have two possibilites: Unknown (if we consider the date that this manuscript was first made public availably in a library or archive for example) or the date the object was made available on the internet (which would be something like 2011).
Of course we could supply "unknown" or "2011", but then in MDS and also in indexing services, publicationYear will always be the date that is most prominently displayed.
For exampled I searched the above mentioned manuscript in the Primo Central catalogue of FU Berlin library and this it what it looks like:
http://primo.kobv.de/primo_library/libweb/action/dlDisplay.do?docId=TN_datacite1599954&loc=adaptor,primo_central_multiple_fe&vid=FUB

My clients want those objects to be found by their original creation date, which I can understand.

Your second point about defining a standard value for unknow mandatory metadata is very interesting and I would very much appreciate if DataCite had a guideline about this.

Best regards
Barbara

[Joan Starr]

Hi Barbara,

The Metadata Working Group has been wrestling with the questions you've raised and we have come to some conclusions. First, we will be providing a list of suggested standard values for use when metadata elements are not known. At present, these are available for the text fields only, however. In the case of the uncertain dates, ones that can be specified only as a date range, we would like to propose the following. It is based on common citation practice:

a. For the PublicationYear property, use the latest year in the range.

b. Then, make use of the Date property to add a date range, with dateType="Issued" or "Created"

Would you be willing to ask your clients if this approach is acceptable to them? We are interested in getting some feedback.

Thank you,

Joan Starr

Metadata Working Group Chair

On Tuesday, October 1, 2013 5:29:12 AM UTC-7, Barbara Hirschmann wrote:

[Barbara Hirschmann]

Hi Joan

the approach you suggest seams reasonable to me, even if I still don't understand what is the actual reason to not allow date ranges in PublicationYear.

However, at the moment I cannot propose your suggestion to my clients because our standard workflow here at ETH is to harvest metadata in Dublin Core format from clients and map it to DataCite in our application. We dont have a mapping routine to individually set the dateType property depending on use case.

Barbara