Discussions towards a Drupal Workshop on Linked Data
Johannes asked to propose topics for a foreseen Drupal Workshop focused on Linked Data.
In my opinion, we should focus on a few essential issues, otherwise we risk being taken away by the million interesting things happening in Drupal.
I would focus on the following crucial points:
1) Integrating Linked Data sets within Drupal both for semantic indexing or as "authority sets" of any kind.
General approach? Queries to the relevant RDF stores and storage of URIs plus perhaps some labels?
Which approach within Drupal? Taxonomies or content types? DERI approach?
URIs as titles of nodes? Otherwise, which label? Exploit node translation for labels in different languages?
How much information / how many relations to import from these sets? Also implement local navigation based on partially imported sets?
2) Exposing contents as RDF:
a) Drupal RDF mappings are not enough, we need more flexibility to comply with different standards.
- How to map CCK content types to more than one class and CCK fields to more than one property?
b) Following the RDF node-arc-node approach, should we exploit Drupal node reference whenever possible instead of other types of fields so that everything is a node (potentially a URI)? How are URIs implemented by Drupal in this scenario?
c) Check the functioning of the Drupal basic SPARQL engine: does it give access to everything? Are all RDF mappings applied?
d) Publishing in Linked Data
Integrating URIs from other datasets within the local content types (point 1) implements one of the basic Linked Data pronciple. How to automate this?
- Burley's experiments
- Calais
- ...
3) How to best visualize and browse triples/relations in Drupal
Most Drupal RDF modules focus on exposure and retrieval of triples and just display them as lists of triples. The Exhibit module "out of the box" doesn't look powerful enough. Learn Exhibit Javascript / CSS to twist it to create better navigation interfaces? Create a new module based on the Java experiments done at IITK?
In addition, perhaps prior to the the workshop we should run a survey, tests and assessment of existing modules, combination of modules and best configurations related to the above points. What is missing? Should we write new modules? If so, can we build on any of the existing ones? What is the status and future of the DERI modules? Is DERI willing to coordinate developments with us?