How the Semantic Web works

A URI identifies any thing
That URI is related to the URI of the document
The document is data about the thing

Tree Vs Web

Tree	Web
OO systems	Semantic Web
Information hiding	Information sharing
Social Hierarchies	Social web
Top-down elaboration	Top-down, Bottom-up & middle-out
Fixed slots	Anyone can say anything about anything
Can't be a web	Can be a tree
Info stored with subject	Can also be stored anywhere else
One ID per object	Things can have many URIs

Put aside Object Oriented and XML paradigms.

URI of Thing to URI of document

Two alternatives

The URI has a hash:
- Eg: http://example.com/products#sku105
- The part on the left of the # is the document URI
The URI has no hash:
- Eg: http://example.com/products/sku105
- This URI must HTTP redirect 303 See Also to the document URI

Do NOT confuse the URI of a thing and the URI of a web page about it.

Each URI identifies one thing only

Hash vs Slash

Hash allows document URI to be generated instantly
Hash works on local file system too
Hash does not work when document would be huge
Useful idiom: http://example.org/product/sku105#it

Criteria for Choosing URIs

Rational classification of the subjects of the data.
Persistence
Persistence
Persistence

Persistence tips

Institutional commitment.
Domain name controlled by those ensuring persistence.
Cleanliness of URI names.
- No mechanisms /cgi-bin/foo.asp?id=x
- No passing attributes /~harriet/latest/
- No internals of file type /foo.rdf
- Technique: Date space
- Technique: Apache rewrite rules
See "Cool URIs don't change"
URN's, DOIs, etc don't help... you need exactly the same...
Institutional commitment. (Try asking W3C)

See also: W3C TAG finding Metadata in URIs, W3C Persistence Policy

Multiple URIs for the same thing

Not the end of the world
Advisable when many agents have data on same thing.
E.G. http://www.w3.org/People/Berners-Lee/card#i and http://www4.wiwiss.fu-berlin.de/dblp/resource/person/100007
Connect using owl:sameAs ('=' in N3)
Store and serve other people's URIs: link!

Be part of the Semantic Web

Persitence - examples: discuss

What to serve: Linked data

Use URIs as names for things
Use HTTP URIs so that people can look up those names.
When someone looks up a URI, provide useful information.
Include links to other URIs. so that they can discover more things.

Ideally: all links out and in
Reasonable size of data returned
Not knowledge bases in zip/tar files!
Link: The use in document d of URI which is in document d'

(See DesignIssues note: Linked Data)

Examples of linked data

Most ontologies
dbview-generated from SQL virtual RDF data
New: D2R Server provides linked data as well as SPARQL
FOAF (home brew, LiveJournal, Opera Community etc)
Semantic wikipedia
Place-names ... etc
The biggest challenge is links to other systems

Lots of data? SPARQL service

SPARQL is the RDF query language
In final stages of standardization
Many implementations
Graph match, optional match,
More concise than equivalent SQL or XQ
Higher conceptual level
More robust against implementation changes

See Jim Melton's W3C Tech Plenary talk ( slides and XTech paper.)

Self-describing Web

"Follow your nose" from URI to what it means.
- Ladder of authority
- Domin owner controls IP address of server
- HTTP URIs have owners who say what they mean
Serve links in both directions

Remember web architecture

Server files with correct Content-type
- eg application/rdf+xml, text/rdf+n3
- Not text/plain or application/binary
Serve links in both directions

Summary

URIs for things, URIs for documents
Pick URIs to be persistent
Serve data about the thing in response to HTTP GET
Use SPARQL for large-scale access.

Semantic Web URIs

Overview

How the web Works