A Semantic Web & Artificial Intelligence

Published in

OpenLink Virtuoso Weblog

5 min readOct 8, 2016

Source: http://rubenverborgh.github.io/WebFundamentals/birds-eye-view/images/scientific-american.jpg

Artificial Intelligence (AI) is once again attracting everyone’s interest. This time around, it’s both connected and disconnected from fundamental ideas behind the seminal “Semantic Web” meme — unleashed in a Scientific American article, circa. 2001.

Here’s a brief definition of some frequently used terms that provide context for this post about reconnecting the notion of a Semantic Web and AI:

Artificial — not human
Intelligence — an ability to apply reasoning and inference to information (that is, data in some context)
Language — systematic use of signs, syntax, and semantics for encoding and decoding information via sentences
Signs — entity identification (Denotation and Connotation Duality)
Syntax — rules for arranging signs to construct a sentence; i.e., syntax
Semantics — meaning associated with each slot occupied by a sign in a sentence
Documents — where sentences are inscribed and persisted
RDF (Resource Description Framework) — language (or framework) for constructing digital sentences that are comprehensible to both humans and machines (courtesy of logic as the system’s conceptual schema)
Logic — formal expression of the fact that everything is related to something else, in a variety of ways; i.e., observation (or data) is a collection of entity relationships categorized (or classified) by relationship type (or relations)

Based on this outline, anyone can create sentences that describe anything, en route to providing information that either a human or machine can decode. Consequently, both humans and machines become smarter as a function of the information at their disposal.

Here’s a collection of statements that describe a variety of entities and how they are related. These sentences convey enough information for a human or machine to make sense of:

Entities
Entity Types (Classes)
Entity Relationship Types (Relations, Attributes, Properties, Connections, Links, Associations, etc.)
Entity Relationships.

RDF-Language Sentences

{_:this <#identifiedBy> <#this> ._:relatedTo <#identifiedBy> <#relatedTo> ._:that <#identifiedBy> <#that> .<#identifiedBy> a rdf:Property .<#identifiedBy> rdfs:domain rdfs:Resource .<#identifiedBy> rdfs:range xsd:anyURI .<#identifiedBy> a owl:ObjectProperty .<#identifiedBy> rdfs:comment "Identfies a Relationship Type [Relation] that associates something identified by indefinite pronoun with an actual Entity Identifier defined by specific Identification Scheme e.g., a Uniform Resource Identifier [URI]." .<#this> <#relatedTo> <#that> .<#this> a rdfs:Resource .<#that> a rdfs:Resource .<#relatedTo> a rdf:Property, owl:ObjectProperty .<#relatedTo> rdfs:comment "Identifies a Relationshipt Type [Relation] that associates two things. Relation Subject and Object Identifiers identify instances of a Resource/Thing/Entity Class i.e., anything to which an Identifier has been assigned." .<#relatedTo> rdfs:domain rdfs:Resource .<#relatedTo> rdfs:range rdfs:Resource .foaf:knows a rdf:Property, owl:ObjectProperty .foaf:knows rdfs:subPropertyOf <#relatedTo> .foaf:knows owl:equivalentProperty schema:knows .foaf:knows rdfs:domain foaf:Person .foaf:knows rdfs:range foaf:Person .schema:knows schema:domainIncludes foaf:Person, schema:Person .schema:knows schema:domainIncludes foaf:Person, schema:Person .foaf:knows rdfs:comment "Identifies a Relationship Type [Relation] that associates one Person with another Person. Relation Subject and Object Identifiers MUST identify instances of a Person Class." .<#johnDoe> foaf:knows <#janeDoe> .<#johnDoe> foaf:name "John Doe" .<#johnDoe> owl:sameAs <#this> .<#johnDoe> foaf:depiction <http://www.johndoe.pro/img/John_Doe.jpg> .<#janeDoe> owl:sameAs <#that> .<#janeDoe> foaf:name "Jane Doe" .<#janeDoe> foaf:depiction <https://pbs.twimg.com/profile_images/535154875059806208/FwaafAh1.png> .}

How a Smart Agent (integrated into existing browsers) perceives the sentences above

First Part of Output from OSDS Browser Extension

Why is this Important?

It solves the following technology-led headaches that continue to bewilder application developers, end-users, decision-makers, investors, organizations, governments, and society at large:

Verifiable Identity — key to preserving individuality
Privacy — applications (agents, apps, actions, modules, libraries, etc.) remain privacy-compromise vectors without verifiable identity controlled by individuals
Data De-silo-fication — applications (agents, apps, actions, modules, libraries, etc.) remain data silo vectors when the items above aren’t addressed properly
Inclusive Innovation — as exemplified by the Web’s original incarnation and fundamental vision.

Artificial Intelligence Impact Example?

Smart Agents driven by AI are expected to be intelligent i.e., consume information and then act on said information with minimal human intervention. In a nutshell, we should be able to communicate with them using natural interaction “best practices” like sentences transmitted via documents, and that shouldn’t be a covert attempt to impose English Language on everyone.

I’ve created two documents containing descriptions of Web Service Actions, a form comprehensible to any Agent (or Bot) that understands subject->predicate->object sentence structure created using RDF-Turtle or JSON-LD notation:

http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/fct-nanotation-description.txt — collection of sentences (using RDF-Turtle notation) that describes Actions associated with a Web Service
http://id.myopenlink.net/DAV/home/KingsleyUyiIdehen/Public/fct-nanotation-jsonld.txt — collection of sentences (using JSON-LD notation) that describes Actions associated with a Web Service

Combined with an ability to lookup terms it might not understand, a Smart Agent should be able to invoke the action described in each of the documents above. This is exactly what our Structured Data Sniffer Browser Extension adds to Chrome, Firefox, Opera, Vivaldi, and the soon to be released update of Edge.

What the Agent sees when it processes RDF-Language sentences created using JSON-LD Notation. Note, you can click your way to immediately experiencing invocation of specific actions.

Same thing, but looking at what the Agent perceives when it processes the same RDF-Language sentences which have been constructed using RDF-Turtle Notation.

Conclusion

Artificial Intelligence, Machine Learning, and whatever comes next, are phrases (recently given new life) that are intimately associated with the very essence of the Semantic Web vision. Ironically, when this vision was initially unleashed, it received a lot of criticism about being too AI oriented.

The great news about the Semantic Web is that, despite its marketing challenges, innovation hasn’t stalled. Thus, awaiting stabilization of marketing memes is a surprising wealth of sophisticated technology that makes Smart Agent technology a reality, and that’s without inadvertently compromising the identity and privacy of its users. :)

A Semantic Web & Artificial Intelligence

Why is this Important?

Artificial Intelligence Impact Example?

Conclusion

Links

Related

Written by Kingsley Uyi Idehen