Making RDF Presentable
Integrated Global and Local Semantic Web Browsing
Lloyd Rutledge, Jacco van Ossenbruggen and Lynda Hardman
CWI, Amsterdam
Overview
-
Browsing Semantics
-
Searching Semantics
- Still first step for access
-
Structuring Semantics
- Document structure as interface to semantics
- Goal
- Any author anywhere adds any semantic statement (triple)
- Any user anywhere accesses it with general semantic browser
- User requests document generated live from triples
Triples: The New Knowledge Unit
-
... vs Document as unit
- Author adds fact once conceived
- URI still browsing unit
- What do you get when you click on a URI on the Semantic Web?
Sesame: Show Every Triple for a URI
Foaf: Show Personal Info for a URI
Browsing Semantics
- Web of hyperlinked URIs, hyperlinks are both
- semantic relationships
- navigational links
-
Common URIs pull triples together into graph
- Any provider anywhere can add one triple at a time
-
Multiple shared vocabularies for each repository
- Who knows what vocabulary you'll stumble upon
Noadster: Generalized Browsing
Current Working Assumptions
-
Triples already harvested
- General architecture still valid for the general Web
-
RDF-annotated media available for display
- Including images, literals and <rdfs:label>
-
Extensible domain-independent foundation
- Reasonable default display for any vocabulary
- "Style sheets" for specific vocabularies
Simile (Haystack, Longwell, Fresnel)
Search in Semantics
- Still first step for access
- In addition to strict URI-based access
- Still necessary for very large stores
- Millions of triples common
- Provides subset of URIs as entry points for browsing
- Like Google, but URIs are entry points into RDF graph
- Flat list of URIs
- Ranking applied to semantics? (see SemRank paper previous session)
- Is current popular clustering enough?
- Here, search is content selection
- Not many complete documents, but components of one new document
Structure in Semantics
- Document structure is most common interface
- One thousand years and going strong
- Document structure on top of semantic structure
- Getting a tree from a graph
- Document retriever becomes document generator
- Get any document you want, just like you want it
- How do you get it?
Clusty: Clustering on Keywords
Clusty Domain-independence
Noadster: Search, Structure and Browse
Specifying Noadster Search and Structure
Topia: Tailored Style and Structure
Topia: Domain-specific Structure Parameters
In case you fell asleep ...
- The Semantic Web is not just AI in XML
- It's more familiar than you think
- It's about writing, sharing and conveying information
- Semantic browsing is hip and happening
- Emerging tools under active development
- An emerging community: development groups and workshops
- Document structure is important interface
- Semantics make search results more sensibly structurable
- Existing clustering tools generate document structure from semantics
- Enticing progress made
- Much exciting work to do
- What's next?
- Good, accessible RDF for browsing
- Stylesheets for appearance and structure: Fresnel
- The Mosaic for Semantic Browsing: Simile/Longwell?
- The CATCH/CHIP project: Personalized Rijksmuseum Collection Access