1. NERC DataGrid Vocabulary Governance Vocabulary Workshop, RAL, February 25, 2009

2. NERC DataGrid Types of Governance  Content Governance  Determines the concepts to be included in controlled vocabularies and their labels (URIs, text terms and definitions)  Technical Governance  Storage, versioning and serving of controlled vocabularies

3. NERC DataGrid Types of Governance  Content Governance  The history of oceanographic data management is littered with examples of poor vocabulary content governance  Decisions made by individuals – even students  Decisions taken with inadequate thought  Definitions conspicuous by their absence  Lazy governance (saying “yes” to all requests)

4. NERC DataGrid Types of Governance  Content Governance  As a consequence we have ‘governed’ international list problems. Consider the so- called ‘ship codes’. We have:  Two hulls with the same URI because they had the same name  Well-known vessels wiped from metadata because they changed name  Inclusion of ‘Helicopter’ and ‘Route: Folkestone - St. Malo/Boulog’ in the vocabulary

5. NERC DataGrid Types of Governance  Content Governance Models  Benign dictator  Can work, but doesn’t scale and what happens if the dictator goes under a bus?  Moderated caucus  Commonly adopted based on e-mail list servers and can work well  Risks associated with the model –Caucus apathy –Caucus intransigence or moderator too weak –Moderator overload (tooling can help) –Moderator too strong  Formal committee  Very slow  Constitution can be a source of problems (no veto!!)

6. NERC DataGrid Types of Governance  Technical Governance  The history of oceanographic data management is littered with examples of poor vocabulary technical governance  Multiple copies –One site served two versions of a list »Both were different »Neither was the master  Weak management –Scrappy files scattered over several FTP sites –Local copies taken –Like Galapagos finches these evolve becoming similar but significantly different  No version labelling or timestamps –Evolving local copies had no idea how they related to their origin

7. NERC DataGrid Types of Governance  Technical Governance Models  Feature Type Catalogue  Designed for GIS world  Working model specified in ISO19135  Equally applicable to vocabularies  NDG Vocabulary Server follows 19135 in outline, but not detail (particularly their terminology)  Ontology Repositories  Ingestions of OWL files  MMI operate one based on Drupal security  Ontology download as files or SPARQL query interface

8. NERC DataGrid Types of Governance  Technical Governance Models  Community distribution sites  Community managed collection resources  Simple catalogue of resources to download as files  OBO is a classic example  Vocabulary server  Terms and lists given URLs  URLs serve XML documents  May also provide API access

9. NERC DataGrid Governance Best Practice  Content Governance Do’s and Don’ts  Do process requests for change as quickly as possible  Do ensure that change requests don’t redefine the collection  Do establish managed deprecation  Do insist on definitions for terms  Don’t change the semantics mapped to a URN  Don’t delete concepts

10. NERC DataGrid Governance Best Practice  Technical governance Do’s and Don’ts  Do provide on-line access to the latest version  Do provide URIs to concepts and collections  Do provide access to previous versions  Do empower content governance with tools  Do provide both clients and service APIs  Don’t allow URNs to change  Don’t make unnecessary change to URLs

11. NERC DataGrid NDG Governance Model  Each list is allocated a content governance authority  These include  BODC internal governance  SeaDataNet Technical Task Team  SeaVoX (Intergovernmental Oceanographic Commission caucus)  ICES  CF Standard Names Committee  NASA Global Change Master Directory  International Hydrographic Bureau  ISO

12. NERC DataGrid NDG Governance Model  Technical governance is delivered by the Vocabulary Server  This has two content update models  Governance authority contacts BODC requesting change (e.g. CF)  BODC monitors governance authority outputs making changes when these change (e.g. GCMD, ISO)  Changes made by BODC staff  This doesn’t scale, so we need a…

13. NERC DataGrid NDG Governance Model  Vocabulary Editor  Governance authority designates individuals who will make change  Authentication identifies these individuals  Authorisation determines which lists in the server portfolio they are allowed to modify  Web-form tool  API to allow edit functionality to be built into other tools  Currently under construction