Transcript Lecture Notes

The Semantic Web
Week 14
Module Website:
http://scom.hud.ac.uk/scomtlm/chs2533
Lecture (SHORT): OWL PIZZAS
Practical (LONGER): Getting to know Protégé-2000
WITH OWL
Recap



Ontologies are a kind of ‘knowledge base’ which
emphasise the capture of domain structure (rather
than eg operational knowledge)
Building knowledge bases involves Knowledge
Engineering – the process of acquisition,
formulation and validation of knowledge
Protégé is a knowledge engineering tool – chiefly
for formulation. Last week we saw how to use it in
its general form. This week you will use it with its
OWL plugin.
This Week..


Lecture: We will look at one well documented OWL
ontology – OWL Pizzas from the University of
Manchester (they’ll be cold by the time they get
here ;-) )
Practical: You will use Protégé with the OWL plugin
and an OWL Pizza tutorial
Recall OWL / Description Logic
We have the open world assumption – things may
be true UNLESS it is specifically stated or can be
inferred that they are not.
So obvious assumptions have to be recorded –
especially ‘disjointness’ between classes

Knowledge Acquisition and Validation
VALIDATION of a formal model is best done in a
‘DIVERSE’ way – using static and dynamic
methods….
The Pizza Application


The requirement is to encode an ontology that
captures the structure of types of Pizzas.The
Protégé tool assists in a kind of method for
knowledge acquisition, and helps eliminate
simple syntax and structural errors
Next week we will see that inference tools are used
in Ontology building
for validation purposes – to check that the ontology
is accurate and self-consistent. Recall that
subsumption is the inference mechanism in DLs.
 to help (re-) structure the class hierarchy.

Knowledge Acquisition without function?
Its difficult to try to capture knowledge without
considering FUNCTION
Eg you have to write a LIST class in a programming
language. How do you do it? By first considering
WHAT FUNCTIONS are required on it!
-checks for null(list)?, operations for adding and
selecting members ETC

For an ontology, we have no strong functional
requirements for it is harder…
Acqusition method for Ontologies in OWL …

Enter classes and subclasses.


Specify disjointness properties between classes



Eg hasIngredient, hasBase, isIngredientOf
Specify characteristics of properties such as functional,
inverse, transitive, symmetric.


eg MeatTopping and VegetarianTopping.
Enter properties and subproperties


eg Pizza, PizzaBase, PizzaTopping, CheeseTopping
eg “isIngredientOf” Is it functional? Transitive? Can it be both?
Does it have an inverse?
Enter Property Domain and Range
Enter Property Restrictions
Practical Work
Open the OWL tutorial PDF document on
http://www.co-ode.org/resources

Work through Chapter 4 (page 19->) of this –
showing you how to enter pizzas into Protégé
NB

there are NO wizards on the Sun implementation
of Protege (there were acting up and had to be
slain by the technicians) so you have to enter
some stuff ‘long hand’
 Protégé is a bit slow and clunky
NEXT week –

we will look at the OWL + the RACER reasoning
system (technology permitting)