Digital Identity is Under Attack: FIDO Paris Seminar.pptx
From Taxonomies to Ontologies
1. From Taxonomies to Ontologies
Christine Connors
Among other things: librarian, information scientist, semantic web advocate and
Founder, TriviumRLG LLC
November 4, 2009
Developing Ontologies, Part Of The Earley & Associates Call Series
2. The Continuum
We are building more complex and powerful data architectures; all types are available
for use on the semantic web
3. Ontology
Thesaurus
Taxonomy
Power
Synonym Ring
List
Folksonomy
Complexity
The Continuum
We are building more complex and powerful data architectures; all types are available
for use on the semantic web
4. The Continuum
Thesaurus
Ambiguity Control
Folksonomy Synonym Ring Synonym Control
Hierarchical Relationships
Personalized Labels Synonym Associative Relationships
Control Scope Note
(Equivalency) (BT, NT, RT, USE, SeeAlso)
Less Complexity More
Taxonomy Ontology
List Ambiguity Control Ambiguity Control
Ambiguity Synonym Control Synonym Control
Control Hierarchical Relationships Hierarchical Relationships
(BT, NT) Associative Relationships
Classes
Properties
Localization
Annotation
Reasoning
“NOT”
Inspired by NISO
Z39.19-2005
5. Terminology
✤ Ontology ~ Given a knowledge domain and scope, the encoding of its concepts, their
properties, and the relationships among them.
✤ Serialization ~ How the ontology is encoded for machine use and transmission. Use what
works for your project: RDF/XML, JSON, N-Triples, whatever!
✤ Triple ~ The basic building block of an ontology; Subject-Predicate-Object.
✤ Graph ~ A visualization of the linked triples.
✤ URI ~ Uniform Resource Indicator, a web-based identifier more generic than the URL.
✤ Namespace ~ A collection of URIs from an authoritative source that share a common identifier.
✤ Qname ~ A shortcut; an abbreviation of the shared namespace identifier, followed by a colon
and a concept name. e.g. dc:creator represents the “creator” element in the Dublin Core
schema. “dc” is defined in the ontology as “http://purl.org/dc/terms/”
7. NT
England
Britain BT
NT
NT BT
BT Wales
Great
Britain NT
NT
BT Scotland
BT
United NT Northern
Kingdom BT Ireland
8. NT
England
Britain BT
God and my right
NT
NT BT
BT Wales
motto Great
Britain NT
NT
BT Scotland
BT
flag
United NT Northern
God Save the Queen anthem Kingdom BT Ireland
official
English language
capital
currency
legislature London
pound sterling
Parliament
9. Transitivity
✤ In a simple hierarchical system (e.g. taxonomy) you have Broader Than/Narrower Than
✤ United Kingdom
✤ Great Britain
✤ Scotland
✤ In an ontology, we can define a Transitive Property (e.g. owl:TransitiveProperty) to cause:
✤ Scotland is a subclass of Great Britain
✤ Great Britain is a subclass of United Kingdom
✤ Therefore, Scotland is a subclass of United Kingdom
10. Symmetry
✤ Sometimes we want to explicitly state that a relationship is bi-
directional.
✤ e.g. “spouse” or “sibling”
Jack Jill
spouse
✤ See Also and Use/Used For conventions are not as complete or as
efficient as a SymmetricProperty.
11. Functional and Inverse
Functional Properties
✤ It can be useful to indicate if a concept can have only ONE value for a
specific attribute.
✤ e.g. a ‘person’ can be EITHER ‘male’ or ‘female’ and not both
✤ It can also be useful to indicate that a value can only be applied to
ONE concept.
✤ e.g. a ‘unique employee id’ can only be assigned to ONE ‘staff
member’
12. Inferencing
✤ It is not necessary in a well-modeled ontology to explicitly encode
every possible triple, many can be inferred.
✤ s: father p: gender o: male
✤ s: father p: typeOf o: parentalRole
✤ s: John p: parentalRole o: father
✤ Therefore
✤ s: John p: gender o: male
13. Things to Remember
✤ Governance ~ even more important due to ontologies being more
complex
✤ BUT you also have better tools to test: SPARQL, inferencing engines &
reasoners
✤ Open-world vs. closed-world assumption
✤ Close it if you must!
✤ Curate the content, not the container
✤ This is more than a descriptive, bibliographic form; you can model the
knowledge, not just the pointers to it
14. There is no “right way.”
There are best practices.
Image by playful.geometer
15. Developing an Ontology
Wednesday November 4th, 1:00 PM ET
Taxonomy Community of Practice Call Series, presented by
Earley & Associates
http://www.earley.com
Thank you
CJMConnors@triviumrlg.com
Nick: CJMConnors at Twitter, Slideshare, LinkedIn, Identi.ca et al
TriviumRLG.com
Notas del editor
Rather than define these here, I’m going to show you some examples. These are some examples you are likely to encounter early on - but are not ALL of the available tools. The most important thing to remember is to take baby-steps. Don’t try to read all of the standards and expect to know how to use them right away! You’ll likely drive yourself mad - it’s a lot to learn, and some things are very different from database and other programming methodologies. Learn each of these things as you encounter a use case for them! And get a good book or two.
This is still the tip of the iceberg!
Why would you want to do this? So that Scotland can inherit properties of its super-classes.
If ‘Jack’ “spouse” ‘Jill’ then ‘Jill’ “spouse” ‘Jack’
You may wonder about the problem of syllogisms, but that is why careful modeling and testing is needed.
Most of what you already know about defining schema and building taxonomies applies to ontology creation as well: know your use case, define your requirements, understand your knowledge domain and the scope of detail you want. Look for existing ontologies to use or buy. Put small pieces together to form your overall model. Make use of subject matter experts, data modeling experts, and keep your core team small.