This document summarizes a presentation on resource and metadata management from a linked data perspective. It discusses projects like Organic.Edunet that deal with heterogeneous metadata from multiple standards and stakeholders. It also outlines the conceptual overview of the EntryStore system for managing linked data, including named graphs, REST API, ACLs, harvesting, querying, and SPARQL. Some lessons learned are around ontologies being difficult for annotators to grasp, the need for clear licensing of data versus metadata, and the complexity involved in practical implementation at large scale.
Spring Boot vs Quarkus the ultimate battle - DevoxxUK
Resource and Metadata Management with a Linked Data perspective
1. Resource and Metadata Management
with a Linked Data perspective
Meetup: Linked Data in Sweden
April 17, 2012
Hannes Ebner
hebner@kth.se
hannes@metasolutions.se
2. Multiple projects with similar goals
• Organic.Edunet as example because of its
characteristics
- Heterogeneous metadata
- Multiple stakeholders and standards
- Ontology
- Big user base
- Used in production environments
- Not typically academic (content-centric)
3.
4.
5. EntryStore – conceptual overview
• Named graphs
• REST API
• ACL
• Harvesting
• Querying
• On-the-fly MD conversions
• SPARQL
• Literal indexing
Open Source
7. Lessons learned - ontology
• Ontology
- Good for domain experts
- Pain for “annotators”
- Nobody used ontology tree
• Flat list with type-ahead is popular
• Difficult to grasp hierarchy
8. Lessons learned - licensing
• Ensure licensing is clear
- Creative Commons
- Open Data Commons
- Custom licenses
- Organic.Edunet: CC0
• Data != Metadata
- Often confused when it comes to licensing
9. Lessons learned - interoperability
• W3C standards everywhere
- Exception: some formats, e.g. IEEE LOM
• Triplification has to be done by experts
• Remix of various vocabularies
• Search engines find resources directly
10. Lessons learned – practical problems
• Complexity
- Try to find a student...
- Big technology stack
• Need for authoritative and quality-controlled
(meta)data
• Legacy data has to be converted and mapped
• Scalability
11. WIN!
No user ever noticed the technology behind
“Rule number 1 in building web software: never show
the URI. If the URI does not have a label, go and beat
somebody up.”
Tim Berners-Lee at LDOW
April 16, 2012
12. FAIL?
OK, there was the ontology...
“The one big ontology approach didn't work”
Tim Berners-Lee at LDOW
April 16, 2012
13. Current work / experiments
• Data-driven Sustainability
- footprinted.org, in collaboration with Sourcemap.com
- Data bridges (EIT ICT Labs)
- Miljöbarometern (visualizations, Google Refine)
- Green Hackathon
• Large scale triplification and interlinking of educational metadata
- ARIADNE
- Open Discovery Space
• Contextualization of cultural heritage metadata
- Europeana (Hack4Europe!)