2 of 9

FAIR Datasets

NDE Partner

collection(s)

Harvester

NDE

dataset

OAI

SPARQL

CLARIAH

FAIRy

Datasets

INEO

Filter & merge

plus

Linked Data

resources

norm

Ingest

facets

Sum-marize

NDE

3 of 9

Facets for FAIR Datasets

Current descriptive set ≅ INEO set + FAIR facets
FAIR Implementation Profiles

Which FAIR facets do we need?

F: PID types/links
F: Metadata formats
F: search engines for (meta)data
A: communication protocol for (meta)data
A: AAI for (meta)data
A: longevity for metadata
I: knowledge representation for (meta)data
I: vocabularies for (meta)data
I: schemas for (meta)data
R: license for (meta)data
R: provenance for (meta)data

Leading to F-A-I-R “stars”, with “enough stars” the dataset can potentially go to INEO

F-UJI

Assess the FAIRness of our datasets?
Technical hints on how to make our FAIRness explicit
How can the registry help the partners repository?

4 of 9

Why do we use CMDI as a pivot format?

CMDI is an XML-based format where you can combine reusable components into resource type specific profiles
It’s in use by CLARIN and 7 of our partners provide their records as CMDI
CMDI can easily morph to mimic other metadata formats, e.g. the DCat of NDE
There is software for a central catalogue, i.e. the VLO, including an ingest an normalization pipeline
Facets can be adapted into the FAIR dataset facets we need
Concept Links provide a bridge to the Linked Data world
Lots of knowledge and tools in our community
CMDI is not a goal/endpoint/result just a (convenient) mean to deal with the heterogeneity in our community

5 of 9

What’s new?

Harvester

Speak SPARQL to the NDE dataset registry to get the records and turn them into CMDI

Resource Summarizer

Find used Schemas and Vocabularies
Also for closed data? e.g. run internally & provide reports for facets (no sensitive data)
On the roadmap for LD, but to be extended for more resource types

Registry endpoints

OAI for the full records
LD for the facets

INEO sideloader

Merge current metadata with static descriptive info, e.g. screenshots
Editor for the descriptive info

6 of 9

FAIR Vocabularies

To be FAIR Datasets need to make explicit their

Knowledge representation
Vocabularies
Schemas
We use a looser LD view on vocabularies, which includes all of these

However, still with a LD focus, which we should try to loose now or in SSHOC.NL

CLARIAH helps to

Find vocabularies recommended by the community

Positive or negative reviews
Crawled from a repository/registry, e.g.
Popularity with known datasets, as assessed by the summarizer

Retain access to vocabulary versions by caching
Access vocabularies in a common way by a common API
Crosswalk between vocabularies by linksets
Alternative hierarchical overlays for vocabularies

We did trials with a local copy of Bartoc filled by crawls from Awesome Humanities and YALC

Now working on a better suitable data model for our own registry

7 of 9

Data Stories

WP4: The 'Spanish' Flu Epidemic in The Netherlands 1918-19

Hosted by DRUID

WP5: 15 years of the popular Dutch chat show 'DWDD' - in data

Hosted by the MediaSuite

Merged their workflows into one

Wrote a spec/schema for the Data Stories, incl. sample instances

Started on the requirements for/design of the editor

Started on open source visualisation plugins, e.g. geo and superset

8 of 9

CLARIAH+ SDR vision

Visualisation module

Jupyter notebook

Data stories editor/ Yasgui

(DANE)

Large Scale Analyses

Authenticated users

Small Scale Analysis

module

Search module

Interactive

Webpage / Enhanced Publication

public git repos

Open datasets

Closed datasets

Open APIs

Closed APIs

User-friendly

FAIR datasets

FAIR tool discovery

Distribution & deployment

Data Stories

Storage/ Archive

runtime

verification