DataONE Architecture and How to Become a Member Node

Room: Pyle Center 335 Expertise Level: Beginner 
Collaboration Area: Information Technology and Interoperability 
Teaser: Overview of the @DataONEorg architecture for info mgrs, grad students, post-docs, faculty, research techs who manage env data

Agenda

Tue, 07/17/2012 - 13:30 to 15:00

Overview of DataONE  (Rebecca Koskela, UNM) (20 mins, 10 mins questions)
 - terminology (CN, MN, ITK)
 - current MNs (chart)
Becoming a Member Node (Matt Jones, UCSB) (40 minutes, 20 minutes discussion)

Tue, 07/17/2012 - 15:30 to 17:00
Identifiers and Packaging in DataONE (Matt Jones) (15/15)

Investigator Toolkit
    DataUp Excel plugin (Carly Strasser) (15/15)
    [Tools Overview (Matt Jones) (15/15)]

Notes

Questions:
What are the criteria for becoming a Member Node?

Is there a difference between EarthCube and DataONE?
There are some overlaps with DataONE - but definitely should be interoperability between the two. D1 playing a technical role in EarthCube. 

How long is D1 funded? 5 years, then reapply for another 5 years

What cyberinfrastructure has D1 built?
The overlay for the MNs to become MNs; CN software; and tools in the ITK

Skipping to "How To:" section

What type of data?  We define environmental data fairly broadly - reference to tag line
Doesn't have to be geo-located

Participants:
UAF- GI & ASF plus the ALISON data (ice data collected by school children)
Arizona 
NGDS - Alabama

Are you collecting information about who is accessing the data? 
Log data about access to datasets - although some MNs allow anonymous access so would just know that someone accessed the dataset - otherwise would know who accesssed it

How is Merritt different that DSpace or Fedora?

How many people are using system? 

How scalable will the system be?  Could it handle a 4 PT dataset?

Identifiers

ESIP did a survey and reccomend:
DUI for datasets
UIDs for granules

If a replication target, then need to be able to accept other MNs identifier system

Why not URL or URI? URI can typically be broken - change of server name will break the
URL or URI.  Is the resolution service reliable? 
Social process of persistence not a technical process

Does D1 count the citations that a dataset receives? Don't track citations ourselves but 
groups like Total Impact are  trying to mainstream dataset citation

Packing
Lack of agreement of what is a dataset
Challenging when talking about data citation
Package is the means for someone to describe what they mean by dataset

In EZID, establishing similar idea of packaging
DataCite has created a new metadata standard with new namespace

Similar to decisions made by NOAA - archive around a packaging concept but have a catalog file describe what's in a package

Sensor streams can cause a problem for some groups because don't want to partition the data

USGS - Water: many projects that generate data that need to go into a institutional repository
Data coming into a database in real time - how should this work?  Discrete snapshots seem
to work - then can assign a unique identifier

D1 might use DOIs - are investigating this now - looking at assigning identifiers to smaller granules. Advice to read the ESIP summary

Actually talking about fine-grained provenance - do you need to have an identifier to define the provenance or is there sufficient information available to identify the granularity?
How to follow data derivations is being worked on by the D1 Provenance & Workflow WG

Implementation is sticking with the coarse grain identification for right now.

Annotation process is closely related to this. 

At what level can data be self-describable? Depends of the metadata that accompanies the data

Will the results of Carly's survey on spreadsheets be shared?  Yes, going to publish and some results already on blog

Model of maximum scientist involvement - what likes best about DataUp

Browse image is important part of the package - especially for educators - Figshare has
a preview of the data
Powerful for discovery

Question about choice of C# for programming language of DataUp - was created by Microsoft - hoping to find a summer student to translate to another programming language

First thing is to get DataUp to communicate with D1 API so then would be able to use any Tier 3 Member Node