1. Knowledge Representation Chapter 7

2. 2 Introduction Real knowledge representation and reasoning systems come in several major varieties. These differ in their intended use, expressivity, features,… Some major families are – Logic programming languages – Theorem provers – Rule-based or production systems – Semantic networks – Frame-based representation languages – Databases (deductive, relational, object-oriented, etc.) – Constraint reasoning systems – Description logics – Bayesian networks – Evidential reasoning

3. 3 Semantic Networks A semantic network is a simple representation scheme that uses a graph of labeled nodes and labeled, directed arcs to encode knowledge. – Usually used to represent static, taxonomic, concept dictionaries Semantic networks are typically used with a special set of accessing procedures that perform “reasoning” – e.g., inheritance of values and relationships Semantic networks were very popular in the ‘60s and ‘70s but are less frequently used today. – Often much less expressive than other KR formalisms The graphical depiction associated with a semantic network is a significant reason for their popularity.

4. 4 Nodes and Arcs Arcs define binary relationships that hold between objects denoted by the nodes. john5 Sue age mother mother(john,sue) age(john,5) wife(sue,max) age(max,34)... 34 age father Max wife husband age

5. 5 Semantic Networks The ISA (is-a) or AKO (a- kind-of) relation is often used to link instances to classes, classes to superclasses Some links (e.g. hasPart) are inherited along ISA paths. The semantics of a semantic net can be relatively informal or very formal – often defined at the implementation level isa Robin Bird Animal RedRusty hasPart Wing

6. 6 Reification Non-binary relationships can be represented by “turning the relationship into an object” This is an example of what logicians call “reification” – reify v : consider an abstract concept to be real We might want to represent the generic give event as a relation involving three things: a giver, a recipient and an object, give(john,mary,book32) give marybook32 john recipient giver object

7. 7 Individuals and Classes Many semantic networks distinguish – nodes representing individuals and those representing classes – the “subclass” relation from the “instance-of” relation subclass instance Robin Bird Animal RedRusty hasPart Wing instance Genus

8. 8 Link types

9. 9 Inference by Inheritance One of the main kinds of reasoning done in a semantic net is the inheritance of values along the subclass and instance links. Semantic networks differ in how they handle the case of inheriting multiple different values. – All possible values are inherited, or – Only the “lowest” value or values are inherited

10. 10 Conflicting inherited values

11. 11 Multiple inheritance A node can have any number of superclasses that contain it, enabling a node to inherit properties from multiple “parent” nodes and their ancestors in the network. These rules are often used to determine inheritance in such “tangled” networks where multiple inheritance is allowed: – If X<A<B and both A and B have property P, then X inherits A’s property. – If X<A and X<B but neither A<B nor B<Z, and A and B have property P with different and inconsistent values, then X does not inherit property P at all.

12. 12 Nixon Diamond This was the classic example circa 1980. Person Republican Person Quaker instance subclass FALSE pacifist TRUE pacifist

13. Building an Ontology

14. Developing an Ontology/Summary There are different methods in the literature The steps are 1.Identify the task. – Identify the range of questions the knowledge base to answer and the kind of facts available for each specific problem instance 2.Assemble the relevant knowledge The is what we call knowledge acquisition process Understand the scope of the knowledge base Collect all relevant literature to the topic Interview domain experts Undertake observation

15. Developing …. 3.Decide the vocabulary of predicates, functions and constants Translate domain level concepts into logic level names The result is the ontology of the domain Ontology means a particular theory of the nature of being or existence Add on existing ontology rather than creating from the scratch 4.Encode general knowledge about the domain – Write down axioms for all the vocabulary terms. This pins down the meaning of terms

16. Building …. Encode the description of the specific problem instance – It involves writing down simple automatic sentences about instances of concepts that are already part of the ontology – Encode general knowledge of the domain – This implies use formal language like first order language or UML or other languages – E.g if two terminals are connected, they have the same signal – Vt 1 t 2 connected (t 1 t 2 ) => signal (t 1 ) = signal (t 2 )

17. Building … 6. Pose queries to the inference procedure and get answers – Let inference procedure operate on the axiom and problem specific facts to derive the facts we are interested in knowing 7. Debug the knowledge base If you miss axioms, some queries may not be answered from the knowledge base Incorrect axioms can be identified because they are false statements about the world V x NumOfLegs (x,4) => Mammals (x) False, reptiles do not have four legs

18. MORE EXAMPLES ON BUILDING AN ONTOLOGY:

19. Example of an ontology for the Semantic Web Flight search engine (e.g. Expedia, Wegolo, Opodo...): – A single interface to query different airline booking systems, compare the results, sometimes aggregate them (mixed airlines complex routes), and perform booking. These websites and the various airlines sites they query communicate without misunderstandings for this task about what are: flights, airports, airlines, departure and arrival points, dates and times, seats, bookings, passengers, fares... They share a common language and knowledge about those “things” Generally, web services need a common ground to achieve interoperability

20. What sort of content should machines share? A vocabulary Facts expressed with this vocabulary Not enough Something is missing to make sure the machines all give the same / a similar meaning to the shared vocabulary, i.e., to constrain its interpretation: Knowledge about the domain, described through the vocabulary More specifically, an ontology expressed in a logical framework NB: Facts can be seen as being part or not of an ontology or a knowledge base, different views can be seen around. Here we will consider that facts are not part of the ontology, but will sometimes display instances on taxonomic graphes.

21. Ontologies An ontology is a “formal, explicit specification of a shared conceptualisation” [Gruber 1993, Borst 1997, Studer et al. 1998] An ontology is a specific artefact: a logical theory and/or a computational object that expresses the intended meaning of a vocabulary, by referring to the nature and the structure of the entities it denotes – cf. Ontology, the philosophical discipline that studies “what there is”, i.e., the nature and structure of reality, and aims at uncovering the Ontology, the theory of reality. ontologies model conceptual knowledge crucial for – Semantic interoperability of systems (through the web or not) – Precise human communication within a scientific or technical domain – Intelligent information extraction, question-answering – Natural language understanding, computational semantics in NLP

22. What is an ontology in practice? A theoretical or computational artefact that 2- formalizes generic, necessary knowledge of this domain and constrains the interpretation of the vocabulary through axioms: – Taxonomic links between classes (“IS-A”, subsumption) e.g., a student is a human: – Student IS-A Human ; ∀ x(Student(x) → Human(x)) – Characterization of the relation arguments, especially the domain and range of binary relations e.g., only students and universities can be related through enrollment: ∀ xy(enrolledAt(x,y) → Student(x) ∧ University(y)) More complex constraints on classes and relations (existence, unicity...) ∀ x(Student(x) → ∃ y(enrolledAt(x,y) ∧ University(y)))

23. Contingent facts and knowledge are not part of the ontology – Facts about specific individuals: Human(Lea), enrolledAt(Lea,UPS) – Contingent knowledge, true in certain circumstances but not necessary: ∀ x(Human(x) → (Student(x) ∨ Teacher(x))).

28. Domain and Range of relations departsFrom and goesTo relate instances of Flight to instances of Airport ∀ xy(departsFrom(x,y) → (Flight(x) ∧ Airport(y))) ∀ xy(goesTo(x,y) → (Flight(x) ∧ Airport(y))) departsOn relates instances of Flight to instances of Date ∀ xy(departsOn(x,y) → (Flight(x) ∧ Date(y))) returnsOn relates instances of RoundTripF to instances of Date ∀ xy(returnsOn(x,y) → (RoundTripF(x) ∧ Date(y))) What can be inferred from these constraints and the taxonomy of classes?

29. Existence and unicity We want to make sure that any flight departs from and goes to some airport, at a given date, etc. We here need to add existence constraints to the domain and range constraints. – ∀ x(Flight(x) → ∃ yzt(departsFrom(x,y) ∧ goesTo(x,z) ∧ departsOn(x,t))) – ∀ x(RoundTripF(x) → ∃ t returnsOn(x,t)) – What about the reverse? Does an airport imply the existence of a flight? In some cases, we have in addition unicity constraints (the binary relation is functional). Here, all the relations are functional, i.e., any flight departs from and goes to a unique airport, on a unique date, etc. – ∀ xyz((departsFrom(x,y) ∧ departsFrom(x,z)) → y = z) – ∀ xyz((goesTo(x,y) ∧ goesTo(x,z)) → y = z) – ∀ xyz((departsOn(x,y) ∧ departsOn(x,z)) → y = z) – ∀ xyz((returnsOn(x,y) ∧ returnsOn(x,z)) → y = z)

30. Non Identity No flight leaves and arrives at the same place ∀ xyz((departsFrom(x,y) ∧ goesTo(x,z)) →¬y = z) Other non-identities?

31. What exactly is a roundtrip flight? A roundtrip flight is composed of two oneway flights with matching airports – New dedicated ternary relation: isComposedOf – Constraint on argument classes (selectional restrictions) ∀ xyz(isComposedOf(x,y,z) → (RoundTripF(x) ∧ OneWayF(y) ∧ OneWayF(z))) – Constraint on departure and arrival airports ∀ xyzab((isComposedOf(x,y,z) ∧ departsFrom(x,a) ∧ goesTo(x,b)) → (departsFrom(y,a) ∧ goesTo(y,b) ∧ departsFrom(z,b) ∧ goesTo(z,a))) – Constraint on dates ∀ xyzab((isComposedOf(x,y,z) ∧ departsOn(x,a) ∧ returnsOn(x,b)) → (departsOn(y,a) ∧ departsOn(z,b))) What constraints are still missing here? – What if we wanted to use a more general relation? (non-direct one- way flights...)

32. Key concepts in ontology Concepts – represent ideas that represent the real world Relation – the interaction between concepts – Two types of relation – Taxonomic relation – sub and super class concept – Associative relation – relate concepts across the structure. E.g Department has student – Relations can be organized into hierarchy hasName – hasFirstName and hasLastName

33. Key terms.. Instance – it the ontology concept with its associated instance – This forms the knowledge base Axioms are used to constrain values for class or instances – Also include more general rules – A person with more than 5 years experience is an expert or a lecturer – VAT is 15%

34. Key concepts Type of ontology – Domain ontology To represent specific domain like education, medicine, agriculture – Upper ontology Used for all discipline You should know what ontology you want to develop

35. Technologies After you define concepts/ontology or list of vocabularies Use tool to manage your ontology 1.Ontology editor – Ontology editors are applications designed to assist in the creation or manipulation of ontologiesontologies – Protégé (Java-based, downloadable, Supports OWL, open source, many sample ontologies, from Stanford University) Protégé – WebODE (Web service offered by the Technical University of Madrid) – Anzo for Excel (Includes an RDFS and OWL ontology editor within Excel; generates ontologies from Excel spreadsheets) – See for a long list of editors -- https://en.wikipedia.org/wiki/Ontology_(information_science) https://en.wikipedia.org/wiki/Ontology_(information_science)

36. Technologies … Learning Ontology learning is the automatic or semi- automatic creation of ontologies, including extracting a domain's terms from natural language text. As building ontologies manually is extremely labor-intensive and time consuming, there is great motivation to automate the process. Information extraction and text mining methods have been explored to automatically link ontologies to documents, e.g. in the context of the BioCreative challenges. [22] [22]

37. Technology Language – An ontology language is a formal language used to encode the ontology.ontology languageformal language – OntoUML is an ontologically well-founded profile of UML for conceptual modeling of domain ontologies. OntoUML – OWL is a language for making ontological statements, developed as a follow-on from RDF and RDFS, as well as earlier ontology language projects including OIL, DAML, andDAML+OIL. OWL is intended to be used over the World Wide Web, and all its elements (classes, properties and individuals) are defined as RDF resources, and identified byURIs. OWLRDFRDFSOILDAMLDAML+OILWorld Wide WebresourcesURIs – Refer for the long list of languages -- https://en.wikipedia.org/wiki/Ontology_(information_science) https://en.wikipedia.org/wiki/Ontology_(information_science)

38. Published examples AURUM - Information Security Ontology, [24] An ontology for information security knowledge sharing, enabling users to collaboratively understand and extend the domain knowledge body. It may serve as a basis for automated information security risk and compliance management. [24] Disease Ontology, [32] designed to facilitate the mapping of diseases and associated conditions to particular medical codes Disease Ontology [32] Gellish English dictionary, an ontology that includes a dictionary and taxonomy that includes an upper ontology and a lower ontology that focusses on industrial and business applications in engineering, technology and procurement Gellish English dictionary WordNet, a lexical reference system WordNet Check for long list of examples -- https://en.wikipedia.org/wiki/Ontology_(information_science) https://en.wikipedia.org/wiki/Ontology_(information_science)