This report is an edited version of the live document created by the participants during the workshop.
- Craig Knoblock
- Ruben Verborgh
- Barry Norton
- Andreas Harth
- Pedro Szekely
- Sebastian Speiser
- Kjetil Kjernsmo
- Jacek Kopecky
- Stefan Scheglmann
The necessity of hypermedia RDF and an approach to achieve it
Paper – Slides
- We want to edit and augment data, not just GET it.
- What's this REST thing?
- Don't make developers look up the spec!
- Is Linked Data inherently RESTful?
- Mike Amundsen's H-factor classification
- What's wrong with the SPARQL Graph Store Spec?
- Jacek: "Hypermedia should be focused on the user." [ and developers should look up the specs ]
- SPARQL graph protocol
- Do we need more expressive linkage rules/templates (rather than links on the resource level only?)
Semi-Automatically Modeling Web APIs to Create Linked APIs
Paper – Slides – Powerpoint
- Automatically find out input and output types and their relationships
- Construct a semantic model based on examples
- Extracting labels
- Extracting relationships is more complicated
- What if the API is not resource-oriented?
- RDFS/OWL to build skeleton using DERI's extension to Google Refine?
- How to get live weather in DBpedia?
- How do we scale up?
The Missing Links
- Mismatch between REST and Semantic Web
- Goal: Build autonomous Web agent to fulfill a task given by people
- Focus on using services to do this
- REST = hypermedia APIs
- Rest can do posts
- How do we get dynamic relations in Semantic Web
- RESTdesc intends to solve this
- Via Erik Wilde: "RFC 5988 very specifically says that they should not be treated as being resolvable."
- Does the REST community need the Semantic Web community?
- Are descriptions the way to go?
- Will it scale?
- How do we describe a service that uploads a photo and returns EXIF metadata differently from a service that only extracts EXIF metadata of a picture that it then forgets? [ thought: make the first a collection ]
Hyperdata: Update APIs for RDF Data Sources
Paper – Slides
- We need a domain-specific API to control write access (new identifiers, security, validation, side effects)
- For RDF, we propose hyperdata.
- Three levels in the data: instance, property, and value
- Resources are a graph that contains predicate indicates that the graph contains info about the resources.
- Statement "templates" can also be contained, using reification. The object is a blank node. It's meant to serve as a service description.
- What if you want inference from the reified statement?
- Service composition
- Service descriptions vs. HATEOAS
- Linked elements/aspects of LAPIs
- Is RDF over HTTP enough?
- Embedded Links vs. Vocabulary vs. Outbound Navigational Links