Implementation Languages Workshop 2022
Date | Time | Meeting Room | Purpose |
---|---|---|---|
Friday 2 December | 9:00-17:00 | Room K319, 1 St Thomas | Requirements Gathering / Pre-discussion |
Monday 5 December | 9:00-17:00 | Room K319, 1 St Thomas | Workshop |
Tuesday 6 December | 9:00-17:00 | Room CS16, 1 St Thomas | Workshop |
Venue: Science Po - Campus 1, Saint-Thomas · 1 Place St Thomas d'Aquin, 75007 Paris, France
Aim
Identification and use of Implementation languages in the DDI Suite of products:
Identify priority implementation languages for DDI products (e.g. RDF, JSON, UML, XML, etc.)
Identify style options for implementation languages
Mappings to produce syntax representations
Moving from conceptual models to serialization
What aspects of implementation should be consistent
Document options, decisions, and reasoning
Provide guidance for variation from the agreed model
based on applied use of product
what needs to be noted and how (need a consistent expression of exceptions and reasons)
Outputs:
Documentation of implementation language decisions
Guidelines for implementing languages in various products
Plan for providing and testing multiple implementation languages for current products`
Background:
The DDI Suite currently expresses its individual products in a number of different implementation languages.
Each product uses one or two of these languages and many are interested in expanding to multiple expressions. Current usage includes XML schema, UML XMI, RDF, and JSON.
The work of the Technical Committee moving the DDI Lifecycle to the COGS production tool will allow us to store content in a CSV format and export to multiple languages. DDI CDI is also working on adding implementation languages beginning with RDF.
Rather than work independently the Technical Committee and the DDI-CDI Working Group believe that it would be beneficial to explore options and provide guidelines to the DDI maintenance and development groups on the use of various features of these languages.
We need to determine if and where we need uniformity and how to inform users of differences in the implementation of individual products in different languages.
This work addresses the approved Scientific Board Work Plan Goals noted in the appendix.
Pre-discussion / Agenda setting (Friday)
Scope of products
DDI-Codebook, DDI-Lifecycle and DDI-CDI
Priority implementation languages
There has been a consensus over several previous discussions that we should have in addition to an XML presentation, RDF/OWL, JSON and UML model, other suggestions have been made. TC proposal for DDI-L is here: TC In person A useful output from this discussion would be grid where for each product a proposal was made for Must, Desirable, Future.
Implementation | DDI-C | DDI-L | DDI-CDI |
---|---|---|---|
XML Instance/Fragment | M | M | M |
JSON /OpenAPI | D | M | M |
JSON-LD |
|
| M |
RDF/OWL | D | M | M |
UML Normative / XMI | D | M | M |
|
|
|
|
Python |
| F |
|
PhP |
| F |
|
C# |
| D |
|
Java |
| F |
|
R |
| F |
|
|
|
|
|
DDI-C backwards compatibility issue - decisions needed to allow other implementations to happen.
OO -
Alignment of implementations across products
Google doc of notes
Common Objects Summary: https://ddi-alliance.atlassian.net/l/cp/DxeYnTpi
Class definitions of major objects
Text overlap with Glossary Group
Relationship to models e.g. ISO 11179, GSIM, Neuchatel
Relationship to implementation standards DC, Schema.org
See slides from Flavio: Mappings across standards.pptx
Identify the ‘major objects’
Work on two examples as POC / template
Concept and its uses
Variable and its uses
Classifications & code lists
Physical & logical data structures
Potential rationalisation of name spaces of ‘objects’
Common representations of comparable patterns, groups of objects, structures examples being
Identification / Definable type / Signification pattern / Collections / Group generic structure / Variable Cascade / Classifications / Value domains
Other use cases
Harmonization between products follows on from 1 & 2 above
Does this make sense, if so in what way ? At what level is there a common structure in terms of basics and how they transfer between products.
A useful outcome of this discussion would be the scope and perhaps which areas would be more of a priority, low hanging fruit and / or specific pain points
What are the implications for different serialisations / implementation languages
Some areas previously raised include:
Nature of implementation languages - relaying features not expressed as explicit objects or specified relationships (links, references) - Model considerations (what is held in model and what is held elsewhere)
How do these vary due to implementation languages
Consistent means of expressing them in various implementation languagesRelationships expressed in XML through hierarchy need to be expressed explicitly in languages like RDF
e.g.
ordering in RDF
constraints SHACL
Understand similarities and differences / are there blockers / unnecessary barriers
DDI-L and DDI-CDI RDF
Other items raised:
There should be a discussion topic to cover the content manager point of view as well as the content user
For discussion:
CDI URLs
Codebook ↔︎ CDI
Lifecycle ↔︎ CDI
CDI content → source metadata
Participants
Name | Organisation |
|
|
Darren Bell | UKDA |
|
|
Pierre-Antoine Champin | University of Lyon |
|
|
Franck Cotton | INSEE |
|
|
Christophe Dzikowski | INSEE |
|
|
Arofan Gregory | CODATA |
|
|
Oliver Hopt | GESIS |
|
|
Jon Johnson | CLOSER |
|
|
Geneviève Michaud | SciencesPo |
|
|
Olof Olsson | SND |
|
|
Flavio Rizzolo (online) | StatsCan |
|
|
Dan Smith | Colectica |
|
|
Romain Tailhurat | INSEE |
|
|
Wendy Thomas | IPUMS |
|
|
Tom Villette | SciencesPo |
|
|