The Semantic Web

Knowledge Graphs

• Industry-scale Knowledge Graphs -- a really nice overview of what Google, Microsoft, Facebook, and others are doing, by the companies themselves
• Enterprise-scale knowledge graphs (PDF) -- the slides from a panel that led to the article above; more visual, less coherent
• What is Google’s Knowledge Graph? -- a readable introduction to what Google's Knowledge Graph is used for and why it matters to web content developers
• schema.org -- the ontology underlying Google's Knowledge Graph and others
  • Datamodel -- notes some impotant features, such as mulitple inheritance is allowed, and multiple inherited property types are disjunctive, i.e., any of the types inherited can be used when filling in the property
  • release notes and ongoing development
• Structured data with Schema.org: The ultimate guide -- how to use schema.org to add semantics to a web page
• Microsoft Concept Graph -- all 85 million IS A relationships in a plain text file! These are text elements not concepts, e.g., look at the example table entry for "apple"

Formalized Web-based Knowledge Representation

For quite a few years now, Tim Berners-Lee, the creator of the world-wide web, and others have tried to create a Semantic Web where web documents include actual machine-usable knowledge in the form of facts, relations, and rules. Recently there has been an uptick in interest and activity, so much so that Franz Inc., the home of AllegroLisp, now have a separate set of commercial products for doing Semantic Web processing. Interestingly, much of the interest is coming from social networking application developers, using the network graph processing aspects of these technologies more than the semantic information aspect.

The core problems in decentralizing knowledge on the web are

• how to represent nested structures and relationships
• how to avoid name conflicts in a decentralized collection of knowwledge

There are two basic answers to representing structure and relationships

• Use a notation that handles nesting, such as XML or JSON
• Use flat lists of "subject - relation - object" triples

The standard answer for name conflicts is to use Internationalized Resource Identifiers (IRIs), i.e., URL syntax and semantics. What worked for millions of documents on 1000s of web servers can work for concepts!

The combination of JSON and IRI's is called JSON-LD. The combination of triples and IRI's is called RDF.

• OWL 2 Primer -- OWL (the Web Ontology Language) has gone through many iterations. It has always existed in various flavors and supported multiple syntaxes. This page introduces the flavors, lets you select which syntax forms examples should be shown in, and gives a small sample knowledge base at the end.
• Owl reasoning examples -- this is from 2003, so some aspects of OWL will have changed, but it shows the kinds of inferences people try to make with OWL.
• Introduction to AllegroGraph: AllegroGraph is Franz Inc.'s commercial Semantic Web library. It is not part of the free Allegro Lisp product, but this introduction is a nice overview of the kinds of data and processing done by a Semantic Web system.
• O'Reilly SPARQL tutorial
• AllegroGraph SPARQL learning centr
• a ton of example SPARQL queries to the WikiData Query Service.
• A slideset on the history of SKOS, microdata, RDFa, and microformats -- slides 61, 62 and 63 show the same example in the 3 formats
• Google's Rich Snippets Testing Tool -- check if your HTML page microdata is correct.
• A tool to extract microdata examples into JSON -- works directly from pasted HTML code
• The Semantic Web: An Introduction -- a very nice introduction that gets into concrete details like RDF vs RDF Schema, but still manages to be quite readable.
• What is Google Knowledge Graph? -- an end user perspective
• The W3C Data Activity home page -- the Data activity has subsumed the Semantic Web activity
• The Attempto Parser -- parses simple English sentences is a controlled syntax into logical forms, including OWL
• Semantic Wikis and Disaster Relief Operations, -- a possible practical application to disaster relief.
• The DARPA Agent Markup Language (DAML) Homepage -- lots of overview and introductions as well as detailed technical examples and links to ontology sites
• A nice tutorial on knowledge markup
• SHOE -- one of the early Semantic Web projects
• A list of resources at O'Reilly
• RDF Primer -- RDF is the WWW foundation for the Semantic Web
• Wilbur -- RDF and DAML in Common Lisp