Re: [rda-datamanagplans] AW: [rda-datamanagplans] AW: [rda-datamanagplans] A short...

Hi everyone;
If I can just clarify some things from my perspective as NERC data scientist who works at two of its designated data centres the BADC and NEODC.
Yes "NERC currently ask for two phases of the DMP. The pre-award plan is much lighter on detail and a fuller plan is only requested once the project has been funded.” and I agree with Joy that they give an indication to archive of what is coming its way and to some extent (but not fully) support reproducibility and validation.
However and this is a big however as soon as project has any degree of complexity the real information that allows you to perform data mangement needs to develop elsewhere.
So if we take the resourcing issue discussed previous posts. The indication of storage and compute resources may be indicated in the DMP. The first thing we then need to is send them a questionaire (see attchached) to flesh out requirements, which are then followed up email. Resource are tailored and assigned. Responsibilities are assigned to people on the projects. For example group workspace management , who can assign access, reponsibilities for backing up, high performance transfer or root acces to virtual machines. All of this data management planning happens on internal CEDA systems.
If we consider the provenance of data captured on collaborative projects where input data sets are transfered onto GWS by different people, possibley reprocessed and combined to form new data sets, which are reanlysed to produce new data products. The complexity of these workflows are not captured except on emails documents and spreadsheets. There is a real need for tools to better support this type of planning and the people doing the science.
This isn't a criticsm of the current NERC DMP template , it is a good pragmatic start . Where a template can be universally given to researchers and works well for relatively starightforward projects. However ther is still room for improvent for more complex collaborative projects which use HPC resources and for the archives and resource providers whcih support them. These project are also not a rarity they are bcoming more and more the norm. My point being approaches which serve both ends of the spectrum need to be developed
Esther
________________________________
From: marjan.grootveld=***@***.***-groups.org [marjan.grootveld=***@***.***-groups.org] on behalf of mgrootveld [***@***.***]
Sent: 18 August 2015 15:59
To: Joy Davidson; 'mhemmje'; 'HermanStehouwer'; ***@***.***; 'rduerr'; 'Active Data Management Plans IG'
Cc: 'mlangset'
Subject: [rda-datamanagplans] Re: [rda-datamanagplans] AW: [rda-datamanagplans] AW: [rda-datamanagplans] A short...
"NERC currently ask for two phases of the DMP. The pre-award plan is much lighter on detail and a fuller plan is only requested once the project has been funded.”
This is also the approach taken by the Dutch funding organisations NWO and ZonMw in their respective pilot studies regarding data management: first a brief data section in the grant proposal, and later on a DMP. Both funders regard the DMP as a living document, to be updated and extended during the project. It would make sense to do this – at least – when a project progress report is due, although it would be better to document all major changes in the project plan if and when they occur; for reproducibility and re-use of the data, but possibly also for writing the Methods section.
I strongly agree with Herman’s concern to keep DMP templates manageable and useful to the average researcher, who is usually not an expert on what data management entails. Therefore I hope that they will become integral parts of research proposals, research progress reports, and research publications. After all, data management should be all in a day’s work – of the researcher, their institutions and their academic disciplines. To this end, we should treat them as normal, low profile, et cetera, rather than trying to pack everything in them. A minimal set of reusable, modular, elements that various stakeholders can agree on might be the way to go. The items mentioned in the Horizon2020 Open Research Data Pilot could be a start, and we have included them in the template in the DANS brochure on http://dans.knaw.nl/en/training-consultancy/consultancy/research-data-ma...
Best regards,
Marjan
Dr Marjan Grootveld
Senior policy officer
+31(0)6 12 10 15 14
Skype: mgrootveld1
***@***.***
Data Archiving and Networked Services (DANS)
DANS offers sustainable access to digital research data. Please visit www.dans.knaw.nl for more information and contact details. DANS is an institute of KNAW and NWO.
From: <***@***.***-groups.org> on behalf of Joy Davidson <***@***.***>
Date: dinsdag 18 augustus 2015 14:26
To: 'mhemmje' <***@***.***-hagen.de>, 'HermanStehouwer' <***@***.***>, "***@***.***" <***@***.***>, 'rduerr' <***@***.***>, 'Active Data Management Plans IG' <***@***.***-groups.org>
Cc: 'mlangset' <***@***.***>
Subject: Re: [rda-datamanagplans] AW: [rda-datamanagplans] AW: [rda-datamanagplans] A short...
Hi everyone,
In my opinion, there are two key areas that a DMP can help to support. The first is to help support reproducibility and validation. The second is to outline plans for data sharing. Funders emphasise data sharing in most of their data related policies, so in the DCC we have been encouraging people to consider DMPs as plans to share (is it going to be appropriate? with who? when and how?). In some cases, sharing won’t be feasible but I think funders are looking to the pre-award DMPS to help them get a feel for what data might be available and accessible at the end of the project.
Supporting validation of published finding is something that could only be supported through updating and fleshing out the DMP over the entire research lifecycle as others have noted. I think we need to cover both aspects. NERC currently ask for two phases of the DMP. The pre-award plan is much lighter on detail and a fuller plan is only requested once the project has been funded. This approach may work well and might be something for us to consider.
In the UK, we are fortunate that RCUK outlined in their Common Principles on Research Data that it is acceptable to use public funds to support data management and sharing. However, it is highly unlikely that any funder would increase the award towards the end of the project to provide additional funding for RDM. So, it is crucial that we can get a good idea of what will be needed to support planned RDM and data sharing from the outset and be sure to have some way of resourcing these requirements – either in the HEI or externally. We encourage researchers to complete their pre-award DMP as a means of helping to identify infrastructure requirements and to be able to request sufficient funds to support RDM and data sharing in new grant applications. The UKDA have a great costing tool that helps researchers to consider additional activities and related costs and gives them an idea of when in the lifecycle the effort is most cost-effective. As we don’t have funder sponsored data centres in all disciplines in the UK, HEIs are expected to play a key role in supporting the retention and reuse of research outputs. This will make budgeting a key concern for institutions as well.
All the best,
Joy
From: Matthias.Hemmje=***@***.***-groups.org [mailto:***@***.***-groups.org] On Behalf Of mhemmje
Sent: 18 August 2015 09:26
To: 'HermanStehouwer'; ***@***.***; 'rduerr'; 'Active Data Management Plans IG'
Cc: 'mlangset'
Subject: [rda-datamanagplans] AW: [rda-datamanagplans] AW: [rda-datamanagplans] A short note on DMPs
I agree that requirements towards expected resources are needed.
Regarding complexity, this is correct. However, not every thing has to be done by operative „average“ researchers.
There are different roles involved and a process model regarding these different actors and their responsibilities will help.
Best regards
Matthias
Prof. Dr.-Ing. Matthias L. Hemmje
FernUniversität in Hagen – Fakultät für Mathematik und Informatik – Lehrgebiet Multimedia und Internetanwendungen
Universitätsstrasse 1 – D-58097 Hagen – Germany
Email: ***@***.***-hagen.de – Web: http://www.fernuni-hagen.de/
Phone: +49 (2331) 987-304 – Mobile: +49 (172) 6840262 – Fax: +49 (2331) 987-4487 – Skype: Matthias.Hemmje
Von:herman.stehouwer=***@***.***-groups.org [mailto:***@***.***-groups.org] Im Auftrag von HermanStehouwer
Gesendet: Montag, 17. August 2015 19:51
An: mhemmje <***@***.***-hagen.de>; ***@***.***; 'rduerr' <***@***.***>; 'Active Data Management Plans IG' <***@***.***-groups.org>
Cc: 'mlangset' <***@***.***>
Betreff: Re: [rda-datamanagplans] AW: [rda-datamanagplans] A short note on DMPs
Yes, but with early declaration of expectations wrt. used infrastructure and needed resources.
Some of the parties involved (at least STFC, but I am sure others as well) do forward planning based on the DMPs for projects in their institutes.
However, overall this is starting to sound quite complex.
How can we keep it manageable and useful to the average researcher?
(after all he/she is usually not an expert on data matters, let alone DMPs, and usually has no desire to become such an expert).
You'll have to excuse me for starting to sound like a broken record.
Cheers,
Herman
On 17/08/15 19:23, mhemmje wrote:
That is why I have said it is multidimensional planning with late binding of resources.
During the proposal the DMP has to comply with the funding framework.
During kick-off and enactement, the early project has to comply with work plan and mangerial structures as finalized in the grant agreement
and ist attachements, as e.g. DOW.
During execution the DMP has to copmly with the available resources which have to be assigned in terms of infrastructure, services,
and human resources.
Best regards
Matthias
Prof. Dr.-Ing. Matthias L. Hemmje
FernUniversität in Hagen – Fakultät für Mathematik und Informatik – Lehrgebiet Multimedia und Internetanwendungen
Universitätsstrasse 1 – D-58097 Hagen – Germany
Email: ***@***.***-hagen.de – Web: http://www.fernuni-hagen.de/
Phone: +49 (2331) 987-304 – Mobile: +49 (172) 6840262 – Fax: +49 (2331) 987-4487 – Skype: Matthias.Hemmje
Von:herman.stehouwer=***@***.***-groups.org [mailto:***@***.***-groups.org] Im Auftrag von HermanStehouwer
Gesendet: Montag, 17. August 2015 11:00
An: ***@***.***; rduerr <***@***.***>; Active Data Management Plans IG <***@***.***-groups.org>
Cc: mlangset <***@***.***>
Betreff: Re: [rda-datamanagplans] A short note on DMPs
Hi David,
As Jamie already noted the demands are not always consistent.
Furthermore, what a researcher needs from a DMP is quite different from what the funder needs.
That is, if a researcher has an overarching DMP for her research it would still have to be adapted for every grant. Which reduces the direct usefulness. (though having one would still be useful and helpful for developing good data praxis).
Cheers,
Herman
On 14/08/15 22:11, ***@***.*** wrote:
That funders seek a DMP does not necessarily mean a one to one mapping with their grant no? Cannot a single DMP be portable and reusable for all funders that request it?
Sent while mobile.
On Aug 14, 2015, at 3:55 PM, rduerr <***@***.***> wrote:
The biggest problem I’ve always seen with even the concept of an active/adaptable DMP is the concept that data sets and projects are related 1 to 1 or maybe many to 1; but not in the many to many fashion which is the way things really work. If I am a researcher who has bene pursuing a line of research for 20 years (e.g., how does volcano plumbing work or what’s going on with the Greenland ice sheet); I may well have a sizable collection of materials that intellectually speaking are one continuous, cohesive collection (e.g., XYZ’s geological study of Antarctica’s Dry Valleys or Joe Blow’s 30 year record of XYZ measurements at Summit Greenland); yet the odds of there only having been one grant and one funding agency involved is probably identically zero. Yes, sure maybe today being able to pursue a single line of research to a meaningful conclusion is more difficult; but I am not convinced that makes the situation better - I think it might actually make this disconnect worse!
Back pre-digital era when that researcher retired and all of their stuff was handed to an archive, it would have been treated as a single collection. When descriptions of it are put on-line now (perhaps involving digitizing some analog materials), they probably would have been split into sub-collections, not by grant but by categories based on scientific utility. For example, in the Antarctic case, Dry Valleys rock samples, Dry Valley’s thin slices; Dry Valley’s chemical assays; etc. In the Greenland case, something like 30 year temperature record at Summit Greenland; 30 year snow albedo Summit, Greenland. Why would anyone want the data split into stuff collected using grant X, stuff collected using grant Y, etc.? Yet that is exactly what this active/adaptive DMP stuff tries to do, which is I think exactly what Herman was saying in the first bullet… What researchers do and what DMP’s aim to do are rather orthogonal at the moment… OK, yes funding agencies might like to see things organized by grant; but that certainly would not make it easy to re-use those data - in fact, organization by grant rather defeats the purpose of maximizing that data’s value.
Now if I had a DMP that actually discussed my line of research that was updated not only during a grant; but to include stuff coming in under any new grants; that might be more realistic.
My 2 cents…
Ruth
On Aug 14, 2015, at 11:19 AM, mlangset <***@***.***> wrote:
A couple of things that I have been thinking about with respect to active/adaptable DMPs:
It would be great if various disciplines could map DMP elements to corresponding elements in their metadata standards. That way if the DMP is kept up-to-date, the end product could be a metadata record describing the data. Tools would likely need to be developed with targeted questions for initially writing the DMP, an interface for updating the DMP, and a utility for outputting standard XML metadata records.
With respect to keeping the DMPs updated, the tools could incorporate a schedule tracker. Prior to the start of the project, PIs would input an anticipated schedule for key milestones in the project where deviations from the initial DMP tend to happen. The tool could push notices to the project team asking if certain elements are still accurate and providing an easy way to edit.
--
Full post: https://www.rd-alliance.org/group/active-data-management-plans-ig/post/s...
Manage my subscriptions: https://www.rd-alliance.org/mailinglist
Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/49532
--
Full post: https://www.rd-alliance.org/group/active-data-management-plans-ig/post/s...
Manage my subscriptions: https://www.rd-alliance.org/mailinglist
Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/49532
--
Full post: https://www.rd-alliance.org/group/active-data-management-plans-ig/post/s...
Manage my subscriptions: https://www.rd-alliance.org/mailinglist
Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/49532
--
Dr. ir. Herman Stehouwer
Max Planck Computing and Data Facility (MPCDF)
RDA Secretariat
***@***.*** 0031-619258815
Skype: herman.stehouwer.mpi
--
Full post: https://www.rd-alliance.org/group/active-data-management-plans-ig/post/a...
Manage my subscriptions: https://www.rd-alliance.org/mailinglist
Stop emails for this post: https://www.rd-alliance.org/mailinglist/unsubscribe/49557
--
Dr. ir. Herman Stehouwer
Max Planck Computing and Data Facility (MPCDF)
RDA Secretariat
***@***.*** 0031-619258815
Skype: herman.stehouwer.mpi

CARL Portage, a new service for research data management in Canada, will be
supporting data management plans. Comments by Marjan, Joy, and Esther all
speak about aspects of the direction that Portage will be taking with DMPs.
Specifically, Marjan noted that "*[D]ata management should be all in a
day’s work – of the researcher, their institutions and their academic
disciplines.*" Joy spoke of key areas for data management support, "*The
first is to help support reproducibility and validation. The second is to
outline plans for data sharing.*" Esther mentioned the importance of
resource planning around storage and computation.
Data management occurs across the full research lifecycle and at multiple
organisational levels: at the level of a project or program, of an
institution, and of the broader research community. Ensuring that all
levels are engaged and responsible for data management aligns with Marjan's
observation that data management is part of the daily work of all three.
The Portage perspective is that the design of a DMP needs to incorporate
significant criteria from each of these levels. Joy's note points to
important criteria from funding agencies and institutions. Esther's
comments about determining resourcing requirements indicate information
valuable to service providers. At the researcher level, a DMP should assist
with finding service providers and with making arrangements for the
transfer of data products once a project ends.
It becomes incumbent upon the DMP service provider to articulate their
value proposition to key stakeholders at these organisational levels. The
responsibility of these stakeholders is to state clearly their data
management principles so that others can understand what the stakeholders
hold as valuable.
Rather than viewing a DMP as an operational plan about how discrete data
activities will be carried out, it may be more useful to think of the 'M'
in DMP as "mobilisation," i.e., how the requirements of researchers,
institutional services, and funding agencies are mobilised to support the
data within a specific project. This is not to downplay the role of
managing data in research but rather to shift the focus of a DMP onto the
relationships among stakeholders needed for a project's data. In preparing
a data mobilisation plan, a researcher would identify strategies for
producing, storing, analyzing, sharing, archiving, and re-using data. The
management of detailed activities would flow from the strategies of a
mobilisation plan. In this context, data management operationalises a
mobilisation plan.
Chuck