In this message l provide some background about developments with regard to Research data management at Wageningen University (Netherlands), and then ask a question in this forum, kindly provide your feedback by replying to this post.

• We have started data related activities by promoting archiving in Dutch National data archives and disciplinary archives. Recently a data librarian was appointed to curate datasets, and the library temporarily hired and " embedded"  scientist to discuss with other researchers the reasons  why they choose to deposit or not. Here is a double interview with both of them.
• We found  that datasets came in slowly and one of the reasons might be that data was managed in an unplanned way. Together with the graduate schools we organised a data management planning course for researchers worlking on their PhD,  and now data management plans are mandatory from 1 April 2014 onwards for PhD projects. See for more information. The purpose of the plans is primarily to make research easier, but we hope that it will make it easier as well to get more datasets archived.
• We register all datasets in out output registration system (untel primarily used to register pubications and act as an institutional repository. Click for the datasets in our repositry. We really want to register them there, because it is the basis for scientists' performance metrics (You can see here some of those metrics by clicking and browsing the units and then the scientists names in the left hand column.) Any publication will earn them credits there, but until now not work on a dataset or a database. Registering datasets in the same system as publications makes it possible to give reseacher credits for data related work (if the management decides to do so)
• And now for the problem that we face at the moment. Our output registration system / repository as we use it now assumes a 1:1 relationship publication : dataset (i.e. d dataset underlies a publication). The very first dataset that we deposited was an example of a  n : 1 relationship (a dataset was used for several articles and a PhD thesis) . Since then we have come across many n : n relationships (publication I uses dataset A B and D, publication II uses dataset B, C, D and E, etc.) We have to create some kind of solution now to keep the data librarian happy, and it should be portable to our next generation systems. You are right, Linked Open Data makes sense here.

So we are looking for a vocabulary to encode the different relationships between publications and datasets. Is any of you such a vocabulary, or of groups working at it?

Hugo Besemer