1. SEMANTIC WEB, OWL & PROTÉGÉ Matthew J Wood CS 570 – Topics in Artificial Intelligence Spring 2013

2. 2 Today’s Web Most of today’s Web content is suitable for human consumption Even Web content that is generated automatically from databases is usually presented without the original structural information found in databases Typical Web uses today people’s seeking and making use of information, searching for and getting in touch with other people, reviewing catalogs of online stores and ordering products by filling out forms

3. 3 Keyword-Based Search Engines Current Web activities are not particularly well supported by software tools Except for keyword-based search engines (e.g. Google, AltaVista, Yahoo) The Web would not have been the huge success it was, were it not for search engines

4. 4 Problems of Keyword-Based Search Engines High recall, low precision. Low or no recall Results are highly sensitive to vocabulary Results are single Web pages Human involvement is necessary to interpret and combine results Results of Web searches are not readily accessible by other software tools

5. 5 The Key Problem of Today’s Web The meaning of Web content is not machine- accessible: lack of semantics It is simply difficult to distinguish the meaning between these two sentences: I am a professor of computer science. I am a professor of computer science, you may think. Well,...

6. 6 The Semantic Web Approach Represent Web content in a form that is more easily machine-processable. Use intelligent techniques to take advantage of these representations. The Semantic Web will gradually evolve out of the existing Web, it is not a competition to the current WWW

7. What is Semantic Web?  Semantic web is a term used more specifically to refer to format and technologies that enable it.  It defines the collection, structuring and recovery of linked data that are enabled by the technologies that provide a formal description of concepts, terms, and relationships within a given knowledge domain.

8. What is Semantic Web?  Sematic Web is the extension of the World Wide Web that enables people to share content beyond the boundaries of applications and websites  Semantic web was term coined by Tim Berners- Lee, the inventor of World Wide Web and the director of the World Wide Web Consortium.  He defines it as “a web of data that can be processed directly and indirectly by machines.

9. Why Semantic Web?  Semantic Web is about common formats for integration and combination of data from diverse sources.  Secondly, it is about language for recording the relationship data and the real world objects.

10. Semantic Web Components  Resource Description Framework (RDF)  RDF Schema (RDFS)  Simple Knowledge Organization System (SKOS)  SPARQL, an RDF query language  Notation3 (N3)  N-Triples  Turtle (Terse RDF Triple Language)  Web Ontology Language (OWL)

11. 11 Explicit Metadata This representation is far more easily processable by machines Metadata: data about data Metadata capture part of the meaning of data Semantic Web does not rely on text-based manipulation, but rather on machine- processable metadata

12. 12 On HTML Web content is currently formatted for human readers rather than programs HTML is the predominant language in which Web pages are written (directly or using tools) Vocabulary describes presentation

13. 13 An HTML Example Agilitas Physiotherapy Centre Welcome to the home page of the Agilitas Physiotherapy Centre. Do you feel pain? Have you had an injury? Let our staff Lisa Davenport, Kelly Townsend (our lovely secretary) and Steve Matthews take care of your body and soul. Consultation hours Mon 11am - 7pm Tue 11am - 7pm Wed 3pm - 7pm Thu 11am - 7pm Fri 11am - 3pm But note that we do not offer consultation during the weeks of the State Of Origin games.

14. 14 Problems with HTML Humans have no problem with this Machines (software agents) do: How distinguish therapists from the secretary, How determine exact consultation hours They would have to follow the link to the State Of Origin games to find when they take place.

15. XML (eXtended Markup Language) A better representation of data XML is a flexible text format that is widely used to structure, store, and transport data. XML is different from HTML because it is not about displaying data. In XML (differently from HTML) you create your own tags to annotate data. XML is used to create other languages such as: XHTML, RSS, RDF, OWL, etc. 15

16. An XML Example 16 Everyday Italian Giada De Laurentiis 2005 30.00 Harry Potter J K. Rowling 2005 29.99

17. RDF (Resource Description Framework) RDF: a standard for describing resources on the Web The meaning of data is encoded in sets of triples. Triples are “subject, predicate, object” statements. Each element of a triple is identified by a URI. URIs represent both resources and relations. RDF is written in XML RDF is to Semantic Web what HTML was to the Web. 17 Harry Potter has as author J. K. Rowling.

18. An RDF Example 18 http://en.wikipedia.org/wiki/J._K._Rowling http://en.wikipedia.org/wiki/ Harry_Potter dc:creator <rdf:RDF xmlns:rdf=http://www.w3.org/1999/02/22-rdf-syntax-ns#http://www.w3.org/1999/02/22-rdf-syntax-ns# xmlns:dc=http://purl.org/dc/elements/1.1/>http://purl.org/dc/elements/1.1/ <rdf:Description rdf:about=“http://en.wikipedia.org/wiki/Harry_Potter”>“http://en.wikipedia.org/wiki/Harry_Potter “http://en.wikipedia.org/wiki/J._K._Rowling

19. Ontologies and OWL An ontology is an explicit description of things and their relations. OWL serves to write ontologies for the Web. OWL is written in XML and built on top of RDF. You can think of OWL as an object-oriented language that defines classes, hierarchy of classes, attributes, relations, etc. OWL is designed to support inference (subsumption and classification) OWL is more expressive than RDF. 19

20. 20 Typical Components of Ontologies Terms denote important concepts (classes of objects) of the domain e.g. professors, staff, students, courses, departments Relationships between these terms: typically class hierarchies a class C to be a subclass of another class C' if every object in C is also included in C' e.g. all professors are staff members

21. 21 Further Components of Ontologies Properties: e.g. X teaches Y Value restrictions e.g. only faculty members can teach courses Disjointness statements e.g. faculty and general staff are disjoint Logical relationships between objects e.g. every department must include at least 10 faculty

22. 22 Example of a Class Hierarchy

23. 23 The Role of Ontologies on the Web Ontologies provide a shared understanding of a domain: semantic interoperability overcome differences in terminology mappings between ontologies Ontologies are useful for the organization and navigation of Web sites

24. 24 The Role of Ontologies in Web Search Ontologies are useful for improving the accuracy of Web searches search engines can look for pages that refer to a precise concept in an ontology Web searches can exploit generalization/ specialization information If a query fails to find any relevant documents, the search engine may suggest to the user a more general query. If too many answers are retrieved, the search engine may suggest to the user some specializations.

25. 25 Web Ontology Languages (2) OWL A richer ontology language relations between classes e.g., disjointness cardinality e.g. “exactly one” richer typing of properties characteristics of properties (e.g., symmetry)

26. 26 Software Agents Software agents work autonomously and proactively They evolved out of object oriented and compontent-based programming A personal agent on the Semantic Web will: receive some tasks and preferences from the person seek information from Web sources, communicate with other agents compare information about user requirements and preferences, make certain choices give answers to the user

27. A Semantic Web Primer27 Intelligent Personal Agents

28. 28 Semantic Web Agent Technologies Metadata Identify and extract information from Web sources Ontologies Web searches, interpret retrieved information Communicate with other agents Logic Process retrieved information, draw conclusions

29. 29 Drawbacks of XML XML is a universal metalanguage for defining markup It provides a uniform framework for interchange of data and metadata between applications However, XML does not provide any means of talking about the semantics (meaning) of data E.g., there is no intended meaning associated with the nesting of tags It is up to each application to interpret the nesting.

30. 30 Requirements for Ontology Languages Ontology languages allow users to write explicit, formal conceptualizations of domain models The main requirements are: a well-defined syntax efficient reasoning support a formal semantics sufficient expressive power convenience of expression

31. 31 Tradeoff between Expressive Power and Efficient Reasoning Support The richer the language is, the more inefficient the reasoning support becomes Sometimes it crosses the border of noncomputability We need a compromise: A language supported by reasonably efficient reasoners A language that can express large classes of ontologies and knowledge.

32. 32 Reasoning About Knowledge in Ontology Languages Class membership If x is an instance of a class C, and C is a subclass of D, then we can infer that x is an instance of D Equivalence of classes If class A is equivalent to class B, and class B is equivalent to class C, then A is equivalent to C, too

33. 33 Reasoning About Knowledge in Ontology Languages (2) Consistency X instance of classes A and B, but A and B are disjoint This is an indication of an error in the ontology Classification Certain property-value pairs are a sufficient condition for membership in a class A; if an individual x satisfies such conditions, we can conclude that x must be an instance of A

34. 34 Uses for Reasoning Reasoning support is important for checking the consistency of the ontology and the knowledge checking for unintended relationships between classes automatically classifying instances in classes Checks like the preceding ones are valuable for designing large ontologies, where multiple authors are involved integrating and sharing ontologies from various sources

35. 35 Reasoning Support for OWL Semantics is a prerequisite for reasoning support Formal semantics and reasoning support are usually provided by mapping an ontology language to a known logical formalism using automated reasoners that already exist for those formalisms OWL is (partially) mapped on a description logic, and makes use of reasoners such as FaCT and RACER Description logics are a subset of predicate logic for which efficient reasoning support is possible

36. 36 Three Species of OWL W3C’sWeb Ontology Working Group defined OWL as three different sublanguages: OWL Full OWL DL OWL Lite Each sublanguage geared toward fulfilling different aspects of requirements

37. OWL DL  OWL DL is so named due to its correspondence with description logic.  Description logic (DL) is a family of formal knowledge representation languages. It is more expressive than propositional logic  It models concepts, roles and individuals, and their relationships  It is used in AI for formal reasoning on the concept of application domain known as Terminological knowledge

38. Protege Tutorial

39. What is protege? Protege is a free, open-source platform to construct domain models and knowledge-based applications with ontologies. Ontologies range from taxonomies, classifications, database schemas to fully axiomatized theories. Ontologies are now central to many applications such as scientific knowledge portals, information management and integration systems, electronic commerce and web services

40. What can you do with Protégé? Create a new ontology from scratch Download and extend an existing ontology Export ontologies in a variety of formats OWL RDF XML Schema

41. How it works Objects in the domain are expressed through a series of interrelated classes Class hierarchy is similar to that used by object- oriented languages such as Java Superclasses Subclasses Sibling Classes Ancestor Classes, etc. Heavy reliance on inheritance Unlike Java, Protégé supports multiple inheritance

42. Individuals and Properties Individual domain objects are expressed as class members Class members have object properties that relate them to members of the same class or other classes Also have data properties Expressed by data type – integer, String, etc. If particular data property does not fit a predefined data type it can be entered as text Ex. Members of the class Soccer_Players have ‘has fanpage’ data property that is expressed as a URL

43. Annotations Provide definitions and comments on the ontology and its contents Can provide annotations for the entire ontology, a specific class, a member of a class, etc.

44. How It Works (cont.) Using Protégé in conjunction with GraphViz software allows users to view the ontology as a semantic network Protégé relies on semantic reasoners to implement description logic

45. Semantic Reasoners Inference engine for description logics Inference process is carried out via forward and backward chaining Protégé supports various reasoners such as FaCT++ and HermiT

46. FaCT++ Semantic reasoner used for OWL DL ontologies Implemented in C++ and supported by Protege Converts the KB into an internal representation using various optimization techniques Uses dependency directed backtracking to check satisfiability of the KB

47. Install Protege Go to http://protege.stanford.edu/doc/owl/getting- started.html to download protege (version 3.x)http://protege.stanford.edu/doc/owl/getting- started.html Protege OWL editor is built with the full installation of protege platform. During the install process, choose the “Basic+OWL” option. For more details: http://protege.stanford.edu/doc/owl/getting- started.html

48. Protege There are two main ways of modelling ontologies: Frame-based OWL Each has its own user interface Protege Frames editor: enables users to build and populate ontologies that are frame-based, in accordance with OKBC (Open Knowledge Base Connectivity Protocol). Classes Slots for properties and relationships Instances for class Protege OWL editor: enables users to build ontology for the Semantic Web, in particular to OWL Classes Properties Instances reasoning

49. Building an OWL Ontology E2: Create a new OWL project Start protege File – New Project – OWL/RDF files – Ontology URI (http://www.pizza.com/ontologies/pizza.owl) – OWL DL – Properties View A new empty Protege-OWL project has been created. Save it in your local file as pizza.owl

50. Named Classes Go to OWL Classes tab The empty class tree contains one class called owl:Thing, which is superclass of everything. E3: Create subclasses Pizza, PizzaTopping and PizzaBase. They are subclasses of owl:Thing. Naming convention no special naming convention consistency

51. Disjoint classes E4: How to say that Pizza, PizzaTopping and PizzaBase classes are disjoint. 1.Select the class Pizza 2.Press “add siblings” button on the disjoint classes widget 3.Add PizzaBase and PizzaTopping 4.Select the class PizzaTopping, 5.Add Pizza and PizzaBase to the disjoint class

52. E5: Create group of classes Create ThinAndCrisyBase and DeepPanBase as the subclasses of PizzaBase, and each of them are disjointed. Select PizzaBase, right click the mouse, select “create subclasses” Follow the wizard to create these two disjoint classes. It will save lots of time when there is need to create lots of disjoint classes.

53. E6: Create some subclasses of PizzaTopping Select PizzaTopping, Create subclaesses as MeatTopping, VegetableTopping, CheeseTopping and SeafoodTopping. Make sure that these classes are disjoint to each other. Select the class MeatTopping, Add disjoint subclasses: SpicyBeefTopping, PepperoniTopping, SalamiTopping and HamTopping Select VegetableTopping: Add disjoint subclasses: TomatoTopping, OliveTopping, MushroomTopping, PepperTopping, OnionTopping, CaperTopping

54. E6: Creating disjoint subclasses Select PepperTopping Add disjoint subclasses: RedPepperTopping, GreenPepperTopping, JalapenoPepperTopping Select CheeseTopping Add disjoint subclasses: MozzarellaTopping, ParmezanTopping Select SeafoodTopping Add disjoint subclasses: TunaTopping, AnchovyTopping and PrawnTopping

55. OWL Properties OWL Properties represent relationships between two objects. There are two main properties: Object properties: link object to object datatype properties: link object to XML Schema datatype or rdf:literal OWL has another property – Annotation properties, to be used to add annotation information to classes, individuals, and properties

57. E7: Create an object property Switch to the “Properties” tab, Use “Create Object Property” button to create a new object property. Rename it to hasIngredient

58. E8: Creating sub-properties Select hasIngredient property Add hasTopping and hasBase as the subproperties

59. Inverse Properties Each object property may have a corresponding inverse property. If some property links individual a to individual b, then its inverse property will link individual b to individual a.

60. E9: Create inverse properties Create a new object property called isIngredientOf Press “Set inverse property” button, Select “hasIngredient” Then the inverse relation has been set up. Select hasBase Create the isBaseOf as the inverse property of hasBase isBaseOf is the subproperty of isIngredientOf Select hasTopping create isToppingOf as the inverse property. isToppingOf is the subproperty of isIngredientOf

61. Functional Properties If a property is functional, for a given individual, there can only be at most one individual to be related via this property. For a given domain, range must be unique Functional properties are also known as single valued properties.

62. Inverse Functional Properties If a property is inverse functional, then its inverse property is functional. For a given range, domain must be unique.

63. Functional vs. inverse functional properties FunctionalProperty vs InverseFunctionalProperty domainrangeexample Functional Property For a given domain Range is unique hasFather: A hasFather B, A hasFather C  B=C InverseFunctional Property Domain is unique For a given range hasID: A hasID B, C hasID B  A=C

64. Transitive Properties If a property is transitive, and the property related individual a to individual b, and also individual b to individual c, then we can infer that individual a is related to individual c via property P.

65. Symmetric Properties If a property P is symmetric, and the property relates individual a to individual b, then individual b is also related to individual a via property P.

66. E10: Make the hasIngredient property transitive Select the hasIngredient property Tick the transitive tick box Select the isIngredientOf property, make sure that the transitive tick box is ticked.

67. E11: Make the hasBase property functional Select the hasBase property Tick the “functional” tick box OWL-DL does not allow datatype properties to be transitive, symmetric or have inverse properties.

68. Property domains and ranges Properties link individuals from the domain to individuals from the range. OWL uses domain and range as axioms in reasoning.

69. E12: Specify the range of hasTopping Select hasTopping Press range button Select PizzaTopping Press OK button PizzaTopping should be displayed in the range list. When multiple classes are added to the range, they represent the union of all classes.

70. E13: Specify Pizza as the domain of the hasTopping property Select hasTopping property Press add domain button Select Pizza Press OK Pizza is displayed in the domain list. When multiple classes are added as domain, they represent as the union of these classes.

71. E14: Specify the domain and range for the isToppingOf property Select the isToppingOf property Set the domain of the isToppingOf property to PizzaTopping Set the range of the isToppingOf property to Pizza.

72. E15: Specify the domain and range for the hasBase property and its inverse property isBaseOf Select the hasBase property Specify the domain as Pizza Specify the range as PizzaBase Select the isBaseOf property Specify the domain as PizzaBase Specify the range as Pizza

73. Property restrictions In OWL, properties are used to create restrictions. Restrictions are used to restrict the individuals that belong to a class Three restrictions: Quantifier restrictions Existential quantifier ( ) Universal quantifier ( ) Cardinality restrictions hasValue restrictions

74. E16: Add a restriction to Pizza Add a restriction to Pizza that specifies a Pizza must have a PizzaBase Select Pizza Select Necessary header to create a necessary condition Select create a restriction wizard Select hasBase as restricted property Select someValueFrom as restriction Put PizzaBase into the filler

75. Add a restriction to Pizza

76. E18: Creating different kinds of Pizzas Create a subclass of Pizza called NamedPizza, and a subclass of NamedPizza called MargheritaPizza. Add comment to MargheritaPizza: A pizza that only has Mozarella and Tomato toppings

77. E19: Adding restrictions to MargheritaPizza To specify that MargheritaPizza has at least one MozzarellaTopping. Select MargheritaPizza Go to “Asserted Conditions”, create new restriction. Select someValueFrom Select hasTopping as the property to be restricted. Enter MozzarellaTopping as the filler Press OK button

78. E20: Adding restrictions to MargheritaPizza To specify that MargheritaPizza has at least one TomatoTopping. Select MargheritaPizza Go to “Asserted Conditions”, create new restriction. Select someValueFrom Select hasTopping as the property to be restricted. Enter TomatoTopping as the filler Press OK button

79. E21: Create AmericanPizza Create AmericanPizza with toppings of pepperoni, mozzarella and tomato. Through cloning and modifying the description of MargheritaPizza. Select MargheritaPizza Select create clone Add additional restriction to AmericanaPizza Adding PepperoniTopping Press OK.

80. E22: Create an AmericanHotPizza and a SohoPizza An AmericanHotPizza is almost the same as an AmericanaPizza, but has JalapenoPepperTopping on it. A SohoPizza is almost the same as a MargheritaPizza, but has additional OliveTopping and ParmezanTopping

81. E23: Make subclasses of NamedPizza disjoint from each other Select MargheritaPizza Press “add all siblings” button on the “Disjoints widget” to make the pizzas disjoint from each other.

82. Using a reasoner Ontology described in OWL-DL can be processed by a reasoner. Go to owl—preference, to make sure that OWL-DL is selected. The main services offered by a reasoner is to test whether or not one class is a subclass of another class. By performing such tests on all of the classes, it is possible for a reasoner to compute the inferred ontology class hierarchy. Another reasoning service is consistency checking – to check whether or not it is possible for the class to have any instances. A class is deemed to be inconsistent if it cannot possibly have any instances.

83. Using Racer In order to reason over the ontology in Protege-OWL, a DIG compliant reasoner should be installed and started. In this example, we use Racer, Download at: http://www.racer- systems.com/products/download/index.phtmlhttp://www.racer- systems.com/products/download/index.phtml Double click RacerPro to start Racer.

84. Invoking the reasoner Having started Racer, the ontology can be sent to the reasoner to automatically compute the classification hierarchy, and also check the logical consistency of the ontology. In Protege, the manually constructed class hierarchy is called the asserted hierarchy. The automatically computed by the reasoner is called the inferred hierarchy. Go to OWL – classify taxonomy – to invoke the reasoner If a class has been reclassified, then the class name will appear in a blue color in the inferred hierarchy. Go to OWL – Check consistency – to invoke the reasoner If a class has been found to be inconsistent, it’s icon will be circled in red color. Computing the inferred class hierarchy is also known as classifying the ontology.

85. Invoke the reasoner

86. E24: Inconsistent classes In order to demonstrate the use of the reasoner to detect inconsistencies in the ontology, we will create a class ProbeInconsistentTopping, Which is the subclass of CheeseTopping Select ProbeInconsistentTopping, go to asserted condition to add named classes, select VegetableTopping and then press OK. Go to OWL – check consistency

87. E25: Classify the ontology again To see ProbeInconsistentTopping is inconsistent.

88. E26: Remove the disjoint statement Between CheeseTopping and VegetableTopping to see what happens Select CheeseTopping Go to Disjoint part Select VegetableTopping, right click and “Delete the selected row”. Classify taxonomy The inconsistency no longer exists.

89. E27: Fix the ontology By making CheeseTopping and VegetableTopping disjoint from each other.

90. SPORTS ONTOLOGICAL EXTENSION IN PROTÉGÉ

91. Project #1 Used Protégé to extend a general sports ontology (sptcsem.owl) Focused on adding classes, individuals, and properties related to the game of soccer Used Fact++ for DL queries