dataset citations on GBIF website?

wardappeltans commented 7 years ago

Dear GBIF, @mdoering, @ahahn-gbif

I noticed that the dataset citations on the GBIF website are not based on the citations from the EML, but show auto-generated ones instead?

e.g.

mdoering commented 7 years ago

Dear Ward, I think this is true but I leave this to @ahahn-gbif, @cgendreau, @kbraak and @thomasstjerne to comment as I am not involved in this. See also gbif/portal-feedback#372

Antonarctica commented 7 years ago

Hi Ward Could you clarify the issue with those citations? The format of the auto generated GBIF portal citation is {dataset.authors} ({dataset.pubDate}) {dataset.title}. [Version {dataset.version}]. {organization.title}. {dataset.type} Dataset {dataset.doi}, accessed via GBIF.org on {YYYY-MM-DD}

So the first is

Sirenko B.I., ed. 2001. List of species of free-living invertebrates of Eurasian Arctic seas and adjacent deep waters. In: Explorations of the fauna of the seas. 51(59). St.Petersburg: 1-132.

Conservation of Arctic Flora and Fauna (2016). Arctic benthic invertebrate collection of the Zoological Institute of the Russian Academy of Science. Occurrence Dataset https://doi.org/10.15468/awamb2 accessed via GBIF.org on 2017-10-13.

For me this issue is due to the fact that they ask to cite a paper which in fact is not the same as the dataset. Actually the dataset metadata show that it has more data than the list (it has more recent data up to 2004). So the Sirenko reference to a paper cannot be the correct one for the dataset..

2. Flanders Marine Institute (VLIZ); (2013). Data collected during the expeditions of the e-learning projects Expedition Zeeleeuw and Planet Ocean. https://doi.org/10.14284/4

Flanders Marine Institute (2017). Data collected during the expeditions of the e-learning projects Expedition Zeeleeuw and Planet Ocean. Version 1.1. Occurrence Dataset https://doi.org/10.14284/4 accessed via GBIF.org on 2017-10-13.

So here the date is different because there was an update of the dataset but the baseline information is there. Does that mean you do not want to have the [Version {dataset.version}] as part of the fields used for generations the dataset citation?

wardappeltans commented 7 years ago

There is indeed still a lot of work to clean up many old dataset citations. I propose to add this as a discussion item on the OBIS nodes training in November, and make a proposal on dataset citations for SG-OBIS-7 in March 2018 (I cannot decide on behalf of them).

Daphnisd commented 7 years ago

About 2: a) There actually haven't been any updates of the dataset since 2013. Our IPT regenerates all datasets each month regardless whether the dataset has been changed. 2013 is the correct publication year, if we define publication year as the year the dataset in it's current version became publicly available. After regeneration the version number will be 1.1, so the version number is meaningless in our IPT.

b) "Flanders Marine Institute" in the GBIF citation refers to the publisher, not the data creator (= an institute listed as resource creator in the IPT). I think it's currently impossible to add an institute or multiple institutes as a data creator in the citation in the GBIF portal, while this is currently possible using the citation generator tool in IPT.

wardappeltans commented 7 years ago

2.b: The IPT auto-generation tool only takes the name of the publishing organization, and not the other organizations listed. They will need to be added manually. Not sure if GBIF disregards this when creating the citation on the GBIF portal.

cgendreau commented 7 years ago

see:

Antonarctica commented 6 years ago

Input from GBIF on the issue see also https://github.com/gbif/registry/issues/43 “ All this said, we really started this exercise to deal with the problem of messy citations. OBIS, on the other hand, has a very well managed system of consistently providing citations in an appropriately standardized format. So in your case, we probably overshot our target a bit by pushing GBIF-generated replacements on top and overruling your version. We had so far been discussing a slightly different solution for fixing this: for datasets provided by OBIS [a filter we will still have to define, based on most likely network membership of datasets], use the custom generated citation string directly as provided from source rather than the auto-generated version from IPT metadata. If the DOI is already part of that citation, use as-is; if a DOI is missing, add the DOI generated during registration with GBIF. Would that work equally well for you? And for other OBIS nodes?

Antonarctica commented 6 years ago

I guess the proposed solution could work in order to get through the old formatted ones. Not sure if it is a solution for those were they ask to cite a paper, first example in my comment on October 14th. technically this is not correct but researcher might prefer that. would that be ok for OBIS?

MattBlissett commented 4 years ago

Hi all

Please see https://github.com/gbif/registry/issues/43 for the details, but in short GBIF is now showing the citation text supplied by the publisher (EML element /eml/additionalMetadata/metadata/gbif/citation) where the dataset is in the OBIS network.

iobis / gbif-marine

dataset citations on GBIF website? #12