Knowledge Component 2 Building Knowledge Representation Knowledge Component
Knowledge Component 2: Building Knowledge Representation Knowledge Component 4: Information Storage 2. 5/4. 2 Ontology: An Introduction Raymond Issa, University of Florida, USA Tomasz Arciszewski, George Mason University, USA 1/50
Module Information • Intended audience – Beginners • Key words – Ontology – Knowledge representation – Ontology building and development tools • Authors: – Raymond Issa, University of Florida, USA – Tomasz Arciszewski, George Mason University, USA • Reviewers: – Chimay Anumba, Loughborough University, UK. – Bill O’Brien, TCCIT IC Committee Chair, University of Texas, Austin, USA 2/50
• • • Review Board: – Renate Fruchter, Ex. Com Past Chair, Stanford University, USA – Carlos Caldas, TCCIT DIM Committee Chair, University of Texas Austin, USA – Bill O’Brian, TCCIT IC Committee Chair, University of Texas Austin, USA – Guillermo Salazar, TCCIT Edu Committee Chair, Worcester Polytechnic Institute, USA – William Rasdorf, TCCIT JCCE editor, North Carolina State University, USA – Chimay Anumba, Loughborough University, UK. The ASCE GCEC Officers: – Tomasz Arciszewski, Ex. Com Chair, George Mason University, USA – Ian Smith, Ex. Com Vice-Chair, EPFL, Switzerland – Hani Melhem, Ex. Com Vice-Chair, Kansas State University, USA The ASCE Technical Council on Computing and IT Officers: – Renate Fruchter, Ex. Com Past Chair, Stanford University, USA – Kim Roddis, Ex. Com Chair, George Washington University, USA – Hani Melhem, Ex. Com Vice Chair, University of Florida Gainesville, USA – Raymond Issa, Ex. Com Secretary, Kansas State University, USA – Ian Flood, Ex. Com Member at Large, University of Florida Gainesville, USA – Ian Smith, Ex. Com Member, EU Liaison, EPFL, Switzerland 3/50
Organization • • • What is an ontology? Why do we need ontologies? How to build an ontology? Examples References and reading 4/50
What is an ontology? 5/50
Concept • A concept is an abstract idea, or a symbolic description of a category of entities, interactions, phenomena, or relationships between them. • It can be presented in various forms: – Natural : Language -> English – Formal: Mathematical A={a 1, b 1} – Visual: Drawings 6/50
Concept: a Natural Form • A concept is a short piece of text using known words (concepts) to describe a new concept • Example: a concept of a truss – A structure which: • • • Has at least three members Has prismatic members All members are pinned-connected All members are two-force members All external loading, including reactions, is applied at joints 7/50
Concept: A Mathematical Form A concept is a set of n symbolic attributes (descriptors) with their values uniquely identifying a category of entities, interactions, phenomena, or relationships between them 8/50
Symbolic Attribute • A symbolic attribute (descriptor) describes qualitative, non-numerical features of an entity • Example (truss), symbolic attributes and their feasible values: – – Member type: prismatic, non-prismatic Connection type: pinned, rigid Member loading type: two-force, multi-force Loading application: joints, everywhere 9/50
Numeric Attribute • A numeric attribute (descriptor) describes quantitative, numerical features of an entity • Example (truss), numeric attributes and their feasible values: – Length: 5 feet, 6 feet, 15 feet – Depth: 16 inches, 18 inches, 20 inches – Weight: 20 pounds, 25 pounds, 30 pounds 10/50
Concept of a Truss: Mathematical Form • • A 1, Member type = prismatic A 2, Connection type = pinned A 3, Member loading type = two-force A 4, Loading application = joints 11/50
Concept of a Truss: Visual Form 12/50
Ontology: a philosophical perspective • From Greek: being and writing about • Trying to find out what entities and what types of entities exist • The most fundamental branch of metaphysics: the study of all kinds of things that exist 13/50
Ontology: General Definitions • A conceptualization of a domain • The best structure of concepts from a given domain for effective computation • A combination of definitions and their relationships 14/50
Gruber’s Definition • Tom Gruber, a leading ontology scholar, is a Canadian computer scientist working in the USA at NIST • His definition: – An ontology is an explicit specification of a conceptualization 15/50
Ontology: A knowledge representation perspective Ontology is a knowledge representation in which the terminologies have been structured to capture the concepts being represented precisely enough to be processed and interpreted by people and machines without any ambiguity 16/50
Why do we need ontologies? 17/50
Four Categorizes of Reasons • • Professional Academic Computational Educational 18/50
Professional Reasons • Modern engineering means cooperation within and across disciplines (a structural engineer working with a mechanical engineer) • Infrastructure security forces cooperation of lawyers, security personnel, firemen, engineers, etc. , all speaking various professional languages • Each profession has its own professional vocabulary • Misunderstanding creates friction, delays, affects productivity, and in critical situations is simply dangerous 19/50
Professional Reasons (continued) Professional Tower of Babel costs time, money, and may be dangerous 20/50
Academic Reasons • Knowledge is a system of concepts, relationships, and processes • No knowledge system can be acquired, built, maintained, or used without understanding its concepts • The key to knowledge is concept understanding 21/50
Academic Reasons “Knowledge is power” Ontologies are vital for understanding/knowledge 22/50
Computational Reasons • Modern engineering is knowledge- and computation-based • Only knowledge- and computation-enabled methods, processes, and tools can be used over local networks or over the Internet • Knowledge and computation-based tools must be ontology-based for consistency and integration reasons 23/50
Computational Reasons (continued) “Ontologies are the key to building computational and knowledge-based tools” 24/50
Present Use of Ontologies • For communication: – Between implemented computational systems – Between humans and implemented computational systems • For computational inference: – For internally representing and manipulating plans and planning information – For analyzing the internal structures, algorithms, inputs and outputs of implemented systems in theoretical and conceptual terms • For reuse (and organization) of knowledge – For structuring or organizing libraries or repositories of plans and planning and domain information 25/50
How to build an ontology? 26/50
Purpose and Scope • Identify the purpose and clarify why the ontology is being built • Identify the scope and determine what its intended uses are (i. e. to be reused, shared, used as part of Knowledge Base, etc. ) • Example: building an ontology for drywalls – Purpose: • a consensual ontology for a construction company and drywall providers - The development of a catalog of the available types of drywalls - Scope: - list of relevant materials, - list of different types of the walls according to the designated uses, etc. 27/50
Building the Ontology • Ontology capture (capturing knowledge) Conceptualization 28/50
Intended model, ontology, conceptualizations and domain world 29/50
Building an Ontology • Identify key concepts and relationships in the domain of interest • Produce precise and unambiguous textual definitions for concepts and relationships • Identify the terms that refer to concepts and their relationships • Example: - Drywall -> (Concept) -> Building material consisting of gypsum formed into a flat sheet and sandwiched between two pieces of heavy paper - Drywall -> (Associated terms) –> Wallboard, gypsum board, GWB, and plasterboard - Greenboard -> (Concept) -> kind of Drywall -> Water resistant wallboard that has asphalt added which gives it a brown gypsum core - Concrete backerboard -> (Concept) -> kind of Drywall -> Concrete reinforced with fiberglass that is typically used as the underlayment for ceramic tile 30/50
Language and Knowledge Representation Remember that natural language definitions determine the knowledge representation of an ontology to be developed 31/50
Concept Identification Strategies • Bottom-up: From the most specific concepts to more abstract concepts. • Top-down: Define most abstract concepts first and then define them into more specific concepts • Middle-out: First the core of the basic terms, and then specifying and generalizing them as required 32/50
Coding Committing to the representation of the ontology (e. g. the terms that define the representation of the ontology such as class, relationship, entities etc. ) and then writing the code 33/50
Languages • As a form of knowledge representation, the selection of the language should depend on what is needed in terms of expressiveness capabilities and reasoning • Expressiveness: How concepts are built in terms of attributes, relations, axioms, among other components • Reasoning: Refer to the main features of the inference engine attached to each language (e. g. simple or multiple inheritance, exception) • Example of traditional ontology languages: - Ontolingua (Based on KIF - Knowledge Interchange Format) OKBC (Open Knowledge Base Connectivity) RDF (Resource Description Framework) OIL (Ontology Interchange Language / Ontology Inference Language) OWL (W 3 C Web Ontology) It is intended to share and publish ontologies on the web 34/50
Ontology Tools • • Development tools. Integrated suites that can be used to build a new ontology from scratch. Evaluation tools. Used to evaluate the content of ontologies ant their related technologies. Merging and alignments tools. Used to solve the problem of merging and aligning different ontologies in the same domain. Querying tools and inference engines. Allow querying ontologies easily and performing inferences with them. 35/50
Conceptualizations and Ontology 36/50
Examples 37/50
Ad-hoc Ontology 38/50
Ad-hoc Ontology • Ontologies acquire knowledge about the world and frame that knowledge into categories and add terminology and constrains them with axioms from traditional logic. • The nodes in Figure represent the Concepts. • Figure shows some specializations of the root, represented by Concept 1 • {Functional Areas, Administration and Buildings} are represented by Concepts 2, 3, and 4 respectively. 39/50
Ad-hoc Ontology • The concept Project Manager, Concept 7, has instances {Bill O’Connell, General Manager, New Hall UF} with fixed attributes {Name, Hierarchy, Project In Charge} • The instance Project Manager O’Connell has the values Name=Bill O’Connell, Hierarchy = General Manager, Project in Charge=New Hall UF • A set of relations among concepts is {is a, part of, has}. These relations are generally denoted as Part. Of(Project Manager, Administration) and they describe associated defined relations within classes, inheritance relations, and instances of properties. 40/50
Construction Business Ontology • The nodes in the ontology represent concepts that have levels of specializations from their parent concept. • The relationships among concepts, for example, Isa(CPVC, Plastic Pipe Fittings) among the Concepts 11 and 9, • The ‘is a’ relation corresponds to a semantic link between two concepts • The ‘has’ relation corresponds to a semantic link that constrain the subsumption relation to a directional relation of containment • The relations of the concepts are Subsumption = It is relation of implication which relates similar to following down the more specific to more general concepts hierarchy of a taxonomy. A Taxonomy = Conserve a hierarchy through generalization/ taxonomy is a central component specializations relations of an ontology 41/50
Construction Business Ontology (continued) • Subsumption relation as in a taxonomy ‘Pipes and Tubes’ subsume ‘Plastic Pipes and Fittings’ - The semantic properties of ‘Pipes and Tubes’ subsume the semantic properties of ‘Plastic Pipes and Fittings’ • • Generalization/specialization relationship between roles as in a taxonomy The role ‘has’ expresses = RELATION OF CONTAINMENT – Roles in the simplest case are represent relationships, and are unidirectional. – 1 -1/4” PVC, Plastic Pipe (13) HAS - ‘Crew’ ‘Daily Output’ The role has expresses as the generalization of all the annotated roles. 42/50
Construction Business Ontology (continued) • Generalization/specialization relationship between roles as in a taxonomy • The role ‘instance of’ expresses = the values for each concept that default values are for any attributes 1 -1/4” PVC, Plastic Pipe (13) Instance of ‘Q-1’ ‘ 42’ ‘L. F. ’ ‘ 2. 45’ - There is a direct correspondence of the default values to the annotated roles by following the order. This is possible by applying rules of inferences that any ontology language provides 43/50
Building Openings Ontology • The Figure represents a taxonomy of openings concepts. • The nodes represent concepts within a category at any level. • The Figure shows opening types classified according to their attributes and their functionality. 44/50
Building Openings Ontology • An aluminum metallic screen door, stainless-steel metallic skylight • An ontology is not necessary a hierarchical tree. It can resemble any a-cyclical structure, as it is shown in the Figure. – The child Metallic is subsumed by Door Screen Frame Skylight 45/50
Summary • The course defines the concept of ontologies and why they are needed. In addition the process of building ontologies is explained and several examples are used to illustrate the process. 46/50
References and Readings 47/50
References and Readings • Davies J. N. , Fensel D. and Van Harmelen F. (2003). Towards the Semantic Web: Ontology-Driven Knowledge Management, John Wiley & Sons. • Farquhar A. , Fikes, R. , Pratt W. and Rice J. (1995). “Collaborative Ontology Construction for Information Integration, ” Stanford University WWW Archive, source: http: //www-kslsvc. stanford. edu: 5915/. • Fensel D. (2004). Ontologies – A Silver Bullet for Knowledge Management and Electronic Commerce, Springer. • Gruber T. R. (1993). A translation Approach to Portable Ontology Specification, Journal of Knowledge Acquisition, (5), pp. 199 -220. • Gruninger M. and Lee J. (2002). “Ontology, Applications and Design, ” Communications of the ACM, (45)2, pp. 39 -41. 48/50
References and Readings (continued) • Lubell J. “XML Representation of Process Descriptions”, http: //ats. nist. gov/psl/xml/process-descriptions. html • Mutis I. (2007). "A Conceptual Framework For Interpretation Of Construction Domain Concept Representations, " Ph. D. Dissertation, University of Florida, Gainesville. • Gomez-Perez A. , Corcho O. and Fernandez-Lopez, M. (2005). Ontological Engineering, 1 st Ed. , Springer, London. • Guarino N. (1997). Understanding, building and using ontologies. International Journal Human-Computer Studies, (46), pp. 293 -310. • Guarino N. (1998). Formal Ontology and Information Systems, FOIS’ 98, Trento, Italy, IOS Press. 12 p • Oguejiofor E. , Kicinger R. , Popovici E. , Arciszewski T. , and De. Jong K. (2004). “Intelligent Tutoring Systems: An Ontology Based Approach, ” International Journal of Computing in Architecture, Engineering, and Construction, (2)2. 49/50
References and Readings (continued) • Ugwu O. O. , Anumba C. J. , Thorpe A. (2001). “Ontology Development for Agent-Based Collaborative Design, ” Journal of Engineering Construction and Architectural Management, (8)3, pp. 211 -224. • Uschold, M. , King, M. , Moralee, S. , and Zorgios, Y. (1998). "The Enterprise Ontology. " The Knowledge Engineering Review, 13(1), 31 - 39. • Sowa, J. F. (1999). Knowledge Representation: Logical, Philosophical, and Computational Foundations, Brooks Cole Publishing Co, Pacific Grove, CA. • Staab, S. Studer, R. , (Editors), (2004), “Handbook on Ontologies”, Springer, 1 st edition. • W 3 C (2004). Web Ontology Language (OWL). World Wide Web Consortium (W 3 C). www. w 3. org/. Accessed April 2006. 50/50
- Slides: 50