(4) Statements and Ontologies

• RDF's graph model can represent any kind of structure
  • there is no built-in bias (different from XML's ordered trees)
  • representing bias (such as ordering) in possible but can become awkward
  • some concepts such as semi-structured data can be almost impossible to represent
• Ontologies allow reasoning over existing data
  • given an ontology, more data can be derived from existing data
  • ontologies never invalidate data, they just add more data
  • classical validation needs to be framed as deriving validation data
• Distribution and decentralization are hidden from users
  • the original vision was often framed as the Giant Global Graph (GGG) [http://en.wikipedia.org/wiki/Giant_Global_Graph]
  • the idealized Semantic Web allows reasoning over the GGG
  • reasoning in the presence of conflicting data and adversaries is a hard problem

(5) Triples have no Identity

• Triples are defined as (Subject, Predicate, Object)
  • an assertion is made about a subject (identified by URI or blank)
  • the assertion is of type predicate (identified by URI)
  • the assertion carries the value object (identified by URI or a string)
• The base model has no concept of origin or identity
  • (Barack Obama, has birthplace, Hawaii)
  • (Barack Obama, has birthplace, Kenya)
  • possible differentiation by adding provenance assertions
• Provenance has to be modeled as an application
  • Reification [http://en.wikipedia.org/wiki/Reification_%28computer_science%29] is possible but results in an explosion of triple volume
  • RDF 2.0 [http://www.w3.org/2009/12/rdf-ws/] might incorporate SPARQL's concept of a Named Graph [http://en.wikipedia.org/wiki/Named_graph]
  • the current landscape is based on well-known and trusted sources of RDF data

(7) Linked Data Principles

• Basic Web principles reframed for the Semantic Web [http://www.w3.org/DesignIssues/LinkedData.html]
  • 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, using the standards (RDF*, SPARQL)
  • include links to other URIs, so that they can discover more things
• Star rating based on how well data is linked
  1. available on the web (whatever format), but with an open licence
  2. available as machine-readable structured data (e.g. excel instead of image scan of a table)
  3. as (2) plus non-proprietary format (e.g. CSV instead of excel)
  4. all the above plus, Use open standards from W3C (RDF and SPARQL) to identify things, so that people can point at your stuff
  5. all the above, plus: link your data to other people's data to provide context
• Per definition, only RDF can be Linked Data [http://dret.typepad.com/dretblog/2009/11/the-linked-data-police.html]

(8) The Linked Data Cloud

(9) From AI to BI

• Semantic Web [Semantic Web (1)] is mainly about data and ontologies
• Linked Data [Linked Data (1)] changes the focus to real-world entities
  • ontologies are no longer the main focus of development
  • focus is shifting to linking data and making data discoverable
• Artificial Intelligence (AI) was the starting point for the Semantic Web vision
• Linked Data effectively implements the Business Intelligence (BI) pattern
  • Extracting data from the underlying back-end architectures
  • Transforming them into RDF by using custom-defined mappings
  • Loading them into The Linked Data Cloud [The Linked Data Cloud (1)]