The funder identification problem

[David Wojick]

The idea that the author's identify the funding subprogram is clear. The problem is how to do that in a universally uniform way so the data can be properly aggregated across all authors and articles? (Plus there are cases where the authors may not know.) That is what the FundRef funder taxonomy is trying to do but it is not simple. By the way, at the just happened SSP annual meeting it was announced that CHORUS and SHARE will collaborate on a funder notification system. This may help.

Reconfiguring the agency contract data systems to include subprogram data is probably not feasible. It would be a big job and the agencies are not that interested in access. Their challenge is spending the money. Most consider public access a chore, especially since there is no funding for it, perhaps even a threat because it will lead to new kinds of performance assessment.

David

On May 30, 2014, at 8:26 PM, Stevan Harnad <amsciforum at GMAIL.COM> wrote:

> Adminstrative info for SIGMETRICS (for example unsubscribe): http://web.utk.edu/~gwhitney/sigmetrics.html
> On Fri, May 30, 2014 at 5:58 PM, David Wojick <dwojick at craigellachie.us> wrote:
>  
> Stevan, you seem to think that I am advocating for a publisher based solution but I am not. I am simply trying to find a good program design. In fact I think CHORUS has wandered away from that design, to the point where the funding agencies may not try to use it.
> 
> Well that much is very welcome news!
>  
> But as for the funder identification problem, it is far from trivial. What you describe below does not work. There is at present no simple way to tell which program funded the research that led to a given article. The grant information does not include this data. How then shall it be created (and made uniform)? This is a deep problem.
> 
> David, I can only repeat: The author of the article (at the very least) -- and hence also the author's institution -- know exactly where they got the funding (program, subprogram, grants officer, everything). And they are very eager to credit their funders for their article output, to justify the funding, to get the next grant instalment, and for renewal or new grant applications.
> 
> In the bottom-up option (2), they simply add a metadata tag for the funding's program/subprogram (in addition to metadata tag for the grant-contract-and-number) in the IR along with the article.
> 
> In the perhaps more sensible option (1), the US Federal funders of research create a database that links their own grant-contract-and-numbers with their own program/subprogram identifiers (and then the IR need only tag the grant-contract-and-number).
> 
> No need for publishers to do any of that. And I don't know why you are suggesting it's such a big deal: There are a finite number of US Federal funding agencies, program/subprograms, grant-contract-and-number, and all the information is know and a matter of record. They have the data; they need only systematically integrate it into a database. 
> 
> Alternatively, authors/institutions can do it distributedly.
> 
> In neither case is it a big deal.
> 
> It only appears like a problem if you look at it from the publisher's viewpoint, where authors are only in the habit of crediting their grant-contract-and-number in the acknowledgements sections of their articles, rather than giving fuller source information. 
> 
> But authors and their institutions have the fuller information. And effective convergent funder and institutional mandates (to deposit in the institutional repository) together with simple, systematic compliance monitoring procedures will ensure that all the requisite metadata are deposited with the article.
>  
> Moreover, how this funder identification problem is solved will probably dictate the form of the US Public Access program. 
> 
> Well, we've made some progress: It won't be the publishers' CHORUS doing it. That's a  non-started, and has been all along.
> 
> Now the choice is between top-down, centrally (1) and bottom-up, distributedly (2). 
> 
> Maybe it's easiest to start bottom-up, distributedly (authors/institutions) and then the funders can harvest the pairs of grant-contract-and-number and program/subprogram metadata tags in the the IRs for all of their articles in order to build up a central database of their own.
> 
> (This institutional-deposit/central harvesting (or institutional export) procedure is by far the optimal mechanism for all article metadata in the IRs and also, if desired, for the full-texts.)
> 
> Stevan Harnad
> 
> On May 30, 2014, at 5:07 PM, Stevan Harnad <amsciforum at GMAIL.COM> wrote:
> 
>> Adminstrative info for SIGMETRICS (for example unsubscribe): http://web.utk.edu/~gwhitney/sigmetrics.html
>> On Fri, May 30, 2014 at 2:39 PM, David Wojick <dwojick at craigellachie.us> wrote:
>>  
>> Stevan, what you describe sounds somewhat like the SHARE program, which is one of the emerging proposed mechanisms for the US Public Access program. See my http://scholarlykitchen.sspnet.org/2013/06/26/universities-propose-to-share-federal-funding-based-articles/ from a year ago.
>> 
>> David, "SHARE" is of course much better than CHORUS because it leaves access-provision in the hands of the universities and research institutions rather than journal publishers. And "SHARE" really just means that the IRs do it rather than the publishers.
>> 
>> But I certainly don't agree that the following is any problem whatsoever:
>> 
>> SHARE has its own specific problems, as does CHORUS. But as far as funder identification goes, SHARE and CHORUS share the problem. This is because the grant information that institutions get does not identify the specific funding program. For example, in DOE the grant contract and grant number merely say that the grant is from the Office of Science, which funds about $5 billion a year. Which of their many programs and sub-programs funded a given grant is not specified. So the institution still has to get that information from the author, just as the publisher does. And the resulting data must be uniform across all institutions, just as with all CHORUS publishers.
>> 
>> So the grant-contract-and-number does not specify the program and subprogram. But the specific funder certainly knows what the program and subprogram is. And the author-and-institution certainly knows what the program and subprogram is. So what's the problem that necessitates publishers (of all parties) to become the custodians of these data, which have nothing to do with publishers?
>> 
>> (a) The exact source of the grant can be disambiguated top-down, with the funders linking the grant-contract-and-number to the exact funding source. That seems the most obvious way to do it. Then the IR need merely include the grant-contract-and-number tag in the institutional repository metadata for each funded article.
>> 
>> (b) Or it can be disambiguated bottom-up, with the author-and-institution including an exact funding source tag in addition to the grant-contract-and-number tag in the institutional repository metadata for each funded article.
>> 
>> Either way, the problem is trivial, and certainly no grounds whatsoever to outsource it to publishers!
>> 
>> Moreover, while the publisher knows about the article, but not the specific funder, the institution need know about neither. At present few, if any, institutional repositories have deposit enforcement that would meet federal standards. Most repository programs are voluntary, so an unreliable source of ALL funded articles. And many institutions have no repository, especially federal laboratories. So all in all the institutions have much further to go than the publishers, as far as the federal program goes. Keep in mind that the paradigm federal access program at this point is PubMed Central, which works with publishers, not institutions.
>> 
>> This is why I have been urging for years that deposit should always be convergent, not divergent, with all funders and all institutions mandating institutional deposit, not institution-external deposit. (Deposits can then be automatically exported to or harvested by any institution-external repositories desired, such as PubMed Central.)
>> 
>> (Institutions that don't yet have a repository are just a piece of free software, some disk-space, and some sysad time from having one. Another non-problem.)
>> 
>> Convergent deposit mandates immediately recruit institutions to monitor and ensure full and timely compliance with the funder OA mandate, whether or not the institution has an OA mandate of its own (and in fact it motivates the institution to adopt an OA mandate of its own, for all of its research output, funded and unfunded).
>> 
>> Again, no grounds whatsoever to outsource any of this to publishers.
>> 
>> This is not to say that SHARE cannot win the public access race. In fact it is rumored that NSF may go the SHARE route, while DOE may go the CHORUS route. NSF works solely with universities, unlike DOE. 
>> 
>> Well, let's hope that DOE too sees the wisdom and practicality of taking the convergent route, with institutions instead of entrusting this funder/institution record-keeping function to a third party with a huge conflict of interest (publishers). 
>> 
>> Stevan Harnad
>> 
>> At 12:12 PM 5/30/2014, you wrote:
>>> 
>>> On Thu, May 29, 2014 at 4:20 PM, David Wojick <dwojick at craigellachie.us > wrote:
>>> 
>>> The core challenge in the US Public Access program is to precisely identify the funders of the research that leads to a given journal article. This sounds easy but it can be a difficult and complex process. The US Government is a vast and complex organization, with hundreds of different offices sponsoring research. Moreover, each office can be referred to in many different ways, creating a major name disambiguation problem in the funder data.
>> To repeat:
>>> 
>>> 
>> It is not for publishers to do record-keeping for the government on the articles they publish, dearly as publishers no doubt wish to hold onto this further potential chain of control over research and researchers' work. 
>>> 
>>> 
>> Researchers' own institutions are the natural ones to do this. As I wrote in my previous posting on this very same issue:
>>  
>> All authors have institutions -- either a university or a research institution. Those institutions have a huge stake in ensuring that their researchers comply with their funder requirements (and they already review all grant applications). Institutions are hence the ones in the position to monitor their own researchers' journal article output, ensure that the funder (if any) is specified in the repository metadata for each published article, and, most important of all, ensure that the deposit is done within the required time-frame (see BOAI recommendation above). 
>> 
>> Repository deposits are time-stamped. Researchers can even be asked to deposit the journal's acceptance letter (in closed access) alongside the final refereed draft, for record-keeping and compliance monitoring purposes. The institution can thereby systematically monitor and ensure timely compliance with funder (and institutional) deposit mandates. (The repository software and the Copy Request Button can then handle any allowable publisher embargo periods in a simple, straightforward way --via the Button till the embargo elapses, and then the deposit automatically becomes OA.)
>>  
>>> CHORUS and FundRef are attacking this funder identification problem using a standardized menu of funder names and DOIs. The basic idea is that the submitting author will pick out the standard names of all the offices that contributed to the research that underlies the submitted article. Again this sounds simple but it is not, because building a comprehensive taxonomy of all possible funders is far from simple.
>> 
>>> 
>> Far from simple -- and far from necessary. Authors' own institutions are the ones that are best positioned to stay abreast of the grants that their researchers have received (in fact they already do so), and their repositories can automatically record what resulting articles are published and deposited, and when.
>>> 
>>> To begin with they have elected to build this menu to identify all the funders in the world, not just the US Federal funders. As a result the menu of funders already has six thousands names and it will probably have many thousands more before it stabilizes. The size of the funder list alone thus creates a big discovery problem, because many funders have similar names. 
>> Let's hope that while (all? some?) publishers are spending their time constructing a mega-database of all their authors' potential funders worldwide, institutions will do the much more simple and sensible thing of constructing a database of all their own employees' funding. (Indeed, they do it already; they need only pool this information in their IR metadata (and their CRIS, if they have one).
>>> 
>> 
>>> 
>> Let journal publishers just manage the peer review of the papers submitted to them and stop trying to create a monopoly over everything else. The research community is perfectly capable of doing its own record-keeping, thank you very much!
>>> 
>> 
>>> Then there is the hierarchy problem, especially within the vast US Government complex. Funding offices occur at many different scales, which are arranged within one another in the tree-like organization chart. For example in the US Energy Department there may be five or more layers of funding offices. Saying which layer should be named in the funding data for a given article is not simple. Moreover if offices in different layers are named for different articles, then the resulting data will have to somehow be aggregated by layer in order to be useful. To make matters worse there are also cross cutting programs that involve multiple offices. In short any taxonomy of US Federal funding offices is going to be a complex system, not a simple listing.
>>> 
>> Needless pseudo-complications. The researcher receives a grant. The grant has an agency and number. That agency and number is one of the metadata tags on all articles arising from that funding that are deposited in the IR.
>>> 
>>> Given these complexities it may be better to have an editor name the funders based on the acknowledgements section of the article, rather than presenting the author with a complex taxonomy of possible funders. There seems to be some experimentation in this direction, but it is a labor intensive solution. The question is also whether the resulting data would be accurate enough for agency purposes; given that acknowledgement has been a relatively informal process. There is also the question of when to collect this funder data, given the labor involved. Should it be upon submission or after acceptance?"
>> The above metadata are enough. Let those who wish to harvest it do so, and do with it, as they will. Publishers have nothing to do with any of this.
>>> 
>> 
>>> 
>> And a reminder: We are talking about monitoring and ensuring compliance with Green OA self-archiving mandates here.
>>> 
>> 
>>> 
>> How funders plan to handle billing and documentation for any Gold OA publishing charges that they may be foolish enough to cover out of scarce research money while the money for Fair-Gold OA is still locked into subscription journals is not what we are discussing here (and certainly not my concern.)
>>> 
>> 
>> Stevan Harnad
>> 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.asis.org/pipermail/sigmetrics/attachments/20140531/6a4d77d0/attachment.html>