URIs in semantic web are real URIs
Lookup: G = Web(u)
- URIs for things derefernce to documents with info about things
- http://www.example.com/foo/bar#thing
- Easy: data in http://www.example.com/foo/bar
- http://www.example.com/foo/bar/thing
- Must return 303 and point to doc with data
- "Link" is use of a symbol which is use of URI in other document
Browsable data
One particular protocol
- A graph G is browsable if
dereferencing "u" gives all arcs in G to and from <u>
& attached molecules
- i.e. documents contain all outgoing and incoming links in G
- Requirement doesn't scale globally :-(
- Nice: Anchored SPARQL queries just work :-)
Tabulator: UI
Goal: A Generic, data browser
- Blue dot for URI to be looked up
- Outline view: companct view of web as tree
- Re-center shift-click
- Query by example (SPARQL subset) - click on properties (not values)
- Alpha demo
- Huge to-do list
Target: iTunes, iPhoto, Quicken etc functionality from a generic browser
Tabulator: RDF
Javascript RDF library: fetch, parse, store, query
- Parse RDF (not yet N3!)
- GRDDL
- Smushing on owl:sameAs, [Inverse]FunctionalProperty
- Look up ontological closure on load
- Dereference query terms during query
Need: fast SPARQL server which will do this
Conclusion
Direct RDF publishing is really valuable and important
- Use URIs .. not just blank nodes.
- Support them with a server
- Give info searcher would probably be interested in
- Give back links
- Use µF only so long as you connect with GRDDL