DAML OIL Ontology Tutorial

1
DAMLOIL Ontology Tutorial

• Chris Wroe, Robert Stevens
• (Sean Bechhofer, Carole Goble, Alan Rector, Ian
  Horrocks.)
• University of Manchester

2
3 goals for today

1. Why use DAMLOIL ontologies?
2. What are the design principles?
3. What is the syntax of DAMLOIL?

3
Providing a controlled vocabulary
Applications
Day Location Qty Pizza type
02/01/02 Cambridge Margherita
1

• Prevent the user spelling magherita differently
• Prevent me from adding a type of car to that
  field
• c) Ensure the users and application have the same
  notion of a magherita pizza
• d) Help an application answer queries such as how
  many vegetarian pizzas were sold?

4
Schema integration
Applications
Ontology
Schema 2
Schema 1
Database 1
Database 2
Schema 3
Database 3
5
TAMBIS
Applications
6
Ontologies to support rules
Drug Catalogue
Rules
Atenolol
oral
25mg
IffAntianginalthen...
Patient Record
?
7
Ontologies to support rules
Drug Catalogue
Drug Ontology
Antianginal
B-Blocker
Atenolol
Patient Record
Guideline
Date
Medication
IffAntianginalthen...

?
311299
Atenolol 25mg tabs bd
8
How are ontologies encoded?
3. Multiple hierarchy
2. Tree
spicy pizza
meat pizza
American hot pizza
9
Property driven classification
Principles

• Most taxonomies are hand crafted
• Difficult to maintain when large or intricate
• Become too large to use but too small to capture
  what you want to say
• Now possible to calculate a hierarchy
• Supply the lego building blocks to allow the
  ontology author or user to define classes
• Use reasoning to check consistency and calculate
  classification based on these definitions

10
Descriptions are key

• Classification driven by description
• Effort moved from manual placement of classes in
  a taxonomy to formal definition of classes.
• The same classes can be classified in different
  ways for different purposes

11
Applications can dynamically build descriptions

• Pizza finder

12
Exercise 1

• Specific Tasks
• Use the information in the menus to write
  definitions for margherita, la reine, soho and
  neptune pizzas
• List the concepts and relationships used in the
  definitions
• General questions
• How else could do you describe types of pizza?
• What parts of the description will drive
  classification?
• What kinds of concepts are used in the
  description?

13
Pizza template

• Pizza types are mainly defined in terms of their
  topping ingredients
• Many high level groups of pizza types depends on
  a categorisation of ingredients

14
Exercise 2

• Use card sorting to categorise pizza ingredients.

15
Exercise 2 beginning with OilEd

• Open an ontology (pizza-tutorial-ex-2.daml)
• Set the namespace
• Add classes to an ontology
• (ham, olive, parmesan)
• Add superclasses (parents) to a class
• (meat, fish, vegetable, and cheese ingredient)
• Reason with the ontology (nothing happens)
• Save an ontology

16
Organisation Terms
Principles

• Classification
• A grouping of things into classes
• Instances
• Members of a class
• Hierarchy
• A stratified organisation of things in which
  there is a top and a bottom
• Taxonomy
• A hierarchical classification

17
Taxonomies as sets
Principles
Eukaryota
Primate
Homo sapiens
Chris
18
Disjointness

• In the DAMLOIL world every class is assumed to
  overlap with every other.
• We need to explicitly state something cannot be a
  member of two classes at the same time e.g. man
  and women

19
Exercise 3 adding a disjoint axiom

• Make tuna a subclass of both meat and fish
  ingredient classes.
• Verify the ontology
• Add a disjoint axiom between meat and fish
  ingredient.
• Send the ontology to the reasoner (verify).
• At this point also add disjoint axioms between
  meat, cheese and vegetable ingredient classes.

20
Relationships

• Non taxonomic links
• Relationship type is called a Property
• Properties can be placed in a taxonomy of their
  own

21
Exercise 4

• Add a property has_topping_ingredient
• Add a property has_part
• Make has_part a super property of
  has_topping_ingredient

22
Restrictions

• Describing a class in terms of its relationships
  to other classes
• Restricting class membership to only those
  individuals that posses these criteria.

23
Exercise 5

• Create a restriction relating margherita pizza to
  its topping ingredients mozzarella and tomato

24
Necessary and sufficient criteria

• To be a member I must have these relationships
  (necessary subclassOf)
• If I have these relationships I must be a member
  (necessary and sufficient sameClassAs)

25
Exercise 6

• Load ontology pizza-tutorial-ex-6.daml
• Create a class called cheesy pizza as a
  subclass of pizza which has a topping ingredient
  of cheese ingredient.
• Change to a SameClassAs definition
• Send the ontology to the reasoner (verify).

26
Negation and disjunction

• DAMLOIL allows logical expressions to be used
  anywhere a named class could be used.
• Allows much more precision and flexibility in
  defining classes.

27
Exercise 7

• Write down a definition for a vegetarian pizza
• Create a new class in the ontology called
  vegetarian pizza and represent the paper based
  definition in DAMLOIL
• Send the ontology to the reasoner.

28
Types of relationship

• hasClass members of class A have this
  relationship with some members of class B
• toClass members of class A (if they have this
  relationship) only have this relationship with
  members of class B

29
Exercise 8

• Change restriction type in the vegetarian pizza
  to toClass.
• Send the ontology to the reasoner

30
Open world assumption

• Open world of the web - always assumes someone
  could add some extra information
• Some one could come along and add a statement to
  say margherita pizza has ham on it.
• We have to cap off the description to say
  margherita pizza has these toppings and I know it
  doesnt have any meat ingredients. (It may have
  herbs etc.)

31
Exercise 9

• Cap off the definition of margherita pizza with
  a toClass restriction.
• Send the ontology to the reasoner.
• (Note we can now explicitly say Neptune has no
  cheese ingredient in the same way)
• Load ontology pizza-tutorial-ex-10.daml to see
  more definitions.

32
Transitivity

• Vegetarians are picky!
• Soho pizza isnt classed as vegetarian in the
  menu
• Some cheeses are made with meat products (rennin)
• The ingredient property is transitive. If
  parmesan has a meat ingredient in it any pizza
  with parmesan should be also be classed as
  containing a meat ingredient

33
Exercise 10

• Make the has_part relationship transitive
• Add rennin as a subclass of meat ingredient
• Add a restriction on parmesan to say it has_part
  some rennin.
• Send the ontology to the reasoner.

34
Cardinality

• More precise numerical constraints can be placed
  on relationships.
• It should be read as members of class A are
  related by this property to this many members of
  class B

35
Exercise 11

• Change the definition of cheesy pizza to a pizza
  with at least 2 types of cheese
• Add a new type of pizza called four cheese
  pizza which has exactly four topping ingredients
  of type cheese ingredient
• Use the reasoner
• Add a disjoint axiom between parmesan and
  mozzarella.
• Use the reasoner

36
Domain and range constraints

• Used to specify what classes can be used with
  properties
• May be useful but can have undesirable effects

37
Exercise 12

• For has_topping_ingredient add a domain of pizza
  and a range of ingredient
• Make intentional errors to test the system e.g.
  add ham as a topping ingredient of onion and add
  four cheeses pizza as a topping of margherita
  pizza.
• Use the reasoner
• Examine the superclasses of onion
• Add a disjoint axiom between pizza and ingredient
• Use the reasoner

38
Things not covered

• Inverse properties, symmetrical properties,
  unique properties.
• Subclass and sameClass axioms.
• Use of individuals

39
Current example ontologies

• GONG
• myGrid

40
Current example applications

• Clinergy not DAMLOIL
• PEDro

41
Providing a controlled vocabulary
Applications
Day Location Pizza type
02/01/02 Cambridge Margherita
3. Multiple hierarchy
2. Tree
meat pizza
spicy pizza
American hot pizza
42
(No Transcript)