|
I think what the blog post highlights,
at least for me, is the "moving data around" piece is much more
significant at the local scale: copying data from an instrument
database onto local storage for analysis, copying analysis
products derived from other data sources, putting it somewhere for
long-term storage, while having it clearly identified and labeled
- those seem like such simple things but surprisingly tricky to do
consistently and reliably for the unique circumstances of actual
experimental lab situations.
There were a few folks I have seen at NDS meetings also at the CENDI/NFAIS meeting - Jane Greenberg from Drexel was there and spoke about Dryad (and metadata). Jim Warren from NIST was there also, and one or two others. A lot of the talks were about motivating scientists to be more open with their data - about the incentive structure, etc. About making it easier, focusing on what scientists actually need, helping them to do their jobs. All the federal agencies there have some sort of data policy to promote integrity and reproducibility - requiring "Data Management Plans" in funding proposals, for instance. But what researchers actually do still seems highly variable and generally limited. Two things I learned - there is a federal "Data Reference Model" which all the agencies are supposed to be mindful of in providing data and metadata: http://en.wikipedia.org/wiki/Data_Reference_Model It's supposed to promote interoperability but it doesn't seem to have met with wide success (or perhaps it has? Wasn't clear from presentations there though). The other thing was - there's a new "Data Carpentry" effort to train scientists on being better managers of their data: http://datacarpentry.org - it seems to be just starting up. As with Software Carpentry it looks like it'll have a python/github slant - something to be aware of, perhaps, maybe NDS should participate somehow... ÂÂ Arthur On 11/25/14, 3:24 PM, Matthew Turk wrote:
|