Building arguments on open data
philippe duchesne
phd@highlatitud.es
@pduchesne
OpenBelgium 2018
Louvain-la-Neuve
March 12th, 2018
how annotated open data fragments can rest your case
Open data annotations technologies
Intro
at the service of engaged citizens
engaged citizen
public report
What?
- Explore data portal/catalogs
- Chunk up data and reassemble into a meaningful argument
- Share that argument
- Have others derive work from your findings
Intro
Ultimately...
- goal is to model and share a thought process based on open data
Intro
- has to be
- shareable -> open data
- verifiable -> assessment of authenticity of sources
- reproducible -> standards-based, open description of process
Preamble : Data standards
Tech
as the 3rd star of open data, it should be obvious by now...
but considering the current state of open data, it's better to rub it in
- XLS sheets
- ZIP files
- a link to a webpage that gathers some data
- ....
are not open data standards!
Verifiability & reproduceability
--> capture data lineage
Tech
Towards a Mosaic standard : Metadata standards
- data provenance metadata
- origin
- authenticity
- based on Dublin Core vocabulary
- data processing modelling
- need for data process description vocabulary
- Considered ontologies: OntoDM, DMOP
qualifying annotations
--> per domain controlled vocabularies
Tech
Towards a Mosaic standard : Controlled Vocabularies
- largely based on Open Annotation Model
Tech
Towards a Mosaic standard : Annotation Model
- consolidates URL fragment syntax
- standard modelling language for visualization
Tech
Towards a Mosaic standard : Visualization model
- Vega from Interactive Data Lab
http://idl.cs.washington.edu
Intro
Towards a Mosaic standard
Open Annotation Model
Dublin Core Metadata
Vega viz grammar
+ Domain controlled vocabularies
Process ML
?
Parliament sessions
Augment transcripts with multimedia recordings
Scenarios
Natural disaster report
Scenarios
Annotate official report with supporting data
Political transparency
Commenting a budget declaration
Scenarios
Political accountability
Scenarios
Analyze political behaviour
But also...
Scenarios
- Open Science
- Data Journalism
- ...
ROI : Enriching back the data
What's next
Grounding Data
Annotations model is RDF-based and queriable in SPARQL
--> graph of all annotations constitutes grounding material
for data itself, and gives context and semantics to data
What's next
Grounding Data : integration
What's next
Now imagine all derived works enriching the original catalog ...
philippe duchesne
phd @ highlatitud.es
@ pduchesne
thank you
questions?
this demo material at
http://demo.highlatitud.es/api/mosaics/openbelgium18
.json
.jsonld
.ttl
this presentation at
https://slides.com/philippeduchesne/ob18-annotations
Building arguments on open data
By Philippe Duchesne
