Thanks very much, this is helpful feedback.
On a related note, Harvard-IQSS created a platform called Dataverse (https://dataverse.org/about) around 2007 and one interesting element is that they published a method for hashing datasets. This is done for the purpose of creating a citation
element that can be used to verify that you have downloaded the same data. Passing along in case this is of interest to the group:
http://best-practices.dataverse.org/data-citation/
Best regards,
Jonathan A. Kennedy
Director of Biodiversity Informatics
Harvard University Herbaria,
Department of Organismic and Evolutionary Biology
From: Daniel Noesgaard <dnoesgaard@gbif.org>
Date: Tuesday, February 19, 2019 at 3:22 AM
To: Quentin Groom <quentin.groom@plantentuinmeise.be>, Tim Robertson <trobertson@gbif.org>
Cc: "Kennedy, Jonathan" <jonathan_kennedy@harvard.edu>, "ipt@lists.gbif.org list" <ipt@lists.gbif.org>, helpdesk <helpdesk@gbif.org>
Subject: Re: [IPT] Daily feeds and archive history
I might also add that every download from GBIF.org–be it a single dataset or an aggregate–is archived and given a unique, persistent DOI for citation. And that citations of downloads count against all the datasets that contributed to that
download.
--
Daniel Noesgaard
Science Communications Coordinator
GBIF | Global Biodiversity Information Facility - Secretariat
Universitetsparken 15
DK-2100 Copenhagen, Denmark
W: www.gbif.org
T: +45 35 32 08 74
From: Quentin Groom <quentin.groom@plantentuinmeise.be>
Date: Tuesday, 19 February 2019 at 08.38
To: Tim Robertson <trobertson@gbif.org>
Cc: "Kennedy, Jonathan" <jonathan_kennedy@harvard.edu>, "ipt@lists.gbif.org list" <ipt@lists.gbif.org>, helpdesk <helpdesk@gbif.org>, Daniel Noesgaard <dnoesgaard@gbif.org>
Subject: Re: [IPT] Daily feeds and archive history
While it would be great to have versioned datasets I generally create a snapshot of the data used in a paper and archive this in Zenodo. This gives complete reproducibility without putting extra demands on the data providers. I do however
need to cite the source and the snapshot.
Regards
Quentin
On Mon, 18 Feb 2019, 17:45 Tim Robertson <trobertson@gbif.org wrote:
Hi Jonathan
(adding GBIF helpdesk to the CC)
This is just a quick answer which I expect will result in follow up questions.
In terms of citation, we use a DOI to identify the concept of a dataset, not the specific version. E.g. https://doi.org/10.15468/cup0nk
If you start deleting copies of data (e.g. a background housekeeping task) what will break are links to the downloads in the IPT pages. https://ipt.huh.harvard.edu/ipt/resource?r=huh_all_records&v=1.3
This may or may not be considered a problem for you.
I think others might have contacted you about suggestions for improving the dataset titles being used but if not I would suggest considering correctly formatted titles as they are used in many places (https://www.gbif.org/dataset/4e4f97d2-4670-4b24-b982-261e0a450faf).
I hope this helps as a start,
Tim
From: IPT <ipt-bounces@lists.gbif.org> on behalf of "Kennedy, Jonathan" <jonathan_kennedy@harvard.edu>
Date: Monday, 18 February 2019 at 18.31
To: "ipt@lists.gbif.org" <ipt@lists.gbif.org>
Subject: [IPT] Daily feeds and archive history
Hi All,
I am finishing an upgrade to the Harvard University Herbaria IPT instance and have configured our feeds for daily auto-publish. The HUH has invested in a mass digitization workflow and we are currently creating ~20,000 new vascular records per month (with minimal data), so we do have new records on a daily basis. However, our DwC archives are fairly large (100MB+), so we can’t keep the daily archive history. I am looking for guidance on how it will work with GBIF dataset citation if we do not preserve each daily archive. It seems problematic if a version of our dataset is used and cited but cannot be reconstructed.
Best regards,
Jonathan A. Kennedy
Director of Biodiversity Informatics
Harvard University Herbaria,
Department of Organismic and Evolutionary Biology
_______________________________________________
IPT mailing list
IPT@lists.gbif.org
https://lists.gbif.org/mailman/listinfo/ipt