From Knowledge Contents to Ontologies: An Introduction to Applied Ontology Dott. ALESSANDRO OLTRAMARI oltramari@loa-cnr.it Istituto di Scienze e Tecnologie della Cognizione (ISTC-CNR) Laboratorio di Ontologia Applicata (Trento) http://www.loa-cnr.it
Outline The context The Business of meaning Solving Semantic Ambiguity Semantic Web and Ontologies 1/2 A definition of knowledge From Multimedia to knowledge contents Knowledge base Ontology and ontologies What is Ontology? What is an ontology? Semantic Web and Ontologies 2/2 The Conceptual bridge for Meaning Negotiation Levels of Ontological Precision Ontologies vs. Knowledge bases Different uses of Ontologies Domain ontologies Ontology-driven information systems Examples of ontological distinctions General Examples Real Examples (link to Protégé) Computational ontologies Ontology Languages Syntax Semantics Efficient Reasoning Support Ontology Web Language (OWL) OWL sub-languages Appendix: OWL primitives Class Object Property Data Property Property Restrictions Special Properties Boolean Combinations Enumerations Do you want more? Links References 2
The context The quality of the (digital) world we live by depends on the quality of information we access to, that needs to be selected, filtered, and organized, Nicholas Negroponte, Digital Beings (1995):
The Business of Meaning (1) Trying to engage with too many partners too fast is one of the main reasons that so many online market makers have foundered. The transactions they had viewed as simple and routine actually involved many subtle distinctions in terminology and meaning Harvard Business Review, October 2001 4
The Business of Meaning (2) Lack of technologies and products to dynamically mediate discrepancies in business semantics will limit the adoption of advanced Web services for large public communities whose participants have disparate business processes Gartner Research, February 28, 2002 5
The Business of Meaning (3) XML is only the first step to ensuring that computers can communicate freely. XML is an alphabet for computers and as everyone who travels in Europe knows, knowing the alphabet doesn t mean you can speak Italian or French! Business Week, March 18, 2002 6
Solving Semantic Ambiguity The main challenge for the new generation of web services is solving semantic ambiguity Semantics analyses meanings Standards vocabularies for application domains are not enough, since Implementing standards needs a lot of time and large human resources Standards only partially apply to concrete cases A simple vocabulary does not specify semantics 7
Semantic Web and ontologies (1/2) Semantic Web has to be conceived as an extension of the current version of the Web. Goal: defining and linking data for automation, integration and reuse in different applications. Methodology: developing standard languages and technologies making explicit the intended meaning of data. In this perspective, the realization of the Semantic Web strongly relies on the use of a shared vocabulary for describing data contents, which will be then used by software and human agents in interactive environments. Ontologies, namely theories of distinctions among entities in a given domain, play a central role in the implementation of the Semantic Web: ontologies provide the meaning of terms in a shared vocabulary 8
A definition of knowledge Knowledge is the main outcome of the process of understanding: by means of interacting with the environment, intelligent agents interpret and represent world situations, acting and reacting to preserve their existence and pursue specific goals. Knowledge needs to be represented in order to become a publiclyaccessible resource and to enable communication between distinct agents (meaning negotiation) knowledge contents 9
From Multimedia to Knowledge contents The World Wide Web is constituted by an intertwined network of multimedia (texts, images, audio files, movies, ) Multimedia constitute the information space (substance) Semantic descriptions turn multimedia into knowledge contents The basic idea behind Semantic Web is to provide a semantic structure (form) to the information space, namely to shape and define meaning of those knowledge contents 10
Knowledge base In computer science, the peculiar notion of knowledge used for characterizing contents corresponds to a set terminological statements (T-box) in a knowledge base (i.e. student is a subclass of person). The assertion component (A-box), namely factual knowledge associated to terminology (i.e. John is a person) is not central here. 11
Applied Ontology and ontologies
What is Ontology? The problem of organizing knowledge contents is directly dependent on the incremental growth of web services and technologies. In this context a new field of computer science appeared, Applied Ontology, whose goal is to rigorously characterize terms and concepts used in information systems, aiming at increasing interoperability and semantic transparency. Applied ontology is based on an interdisciplinary approach (philosophy, linguistics, economy, social science, etc.), where formal logics plays a unifying role. 13
What is an ontology? An artifact designed with the purpose of expressing the intended meaning of a (shared) vocabulary A shared vocabulary plus a specification (characterization) of its intended meaning An ontology is a specification of a conceptualization [Gruber 95] i.e., an ontology accounts for the commitment of a language* to a certain conceptualization * Natural or Formal Language 14
Guarino (1998): what is an ontology? Commitment Conceptualization Language Interpretation Models Intended models Ontology 15
Oltramari (2006): Revisiting Guarino 16
Levels of Ontological Precision a taxonomy A logical theory a glossary a thesaurus a DB/OO scheme Ontological precision 17
Different uses of ontologies (Processing time) Simple semantic access Lightweight ontologies define a minimal terminological structure with respect to specific community/domain (i.e. tourism). Intended meaning of terms known in advance Lightweight ontologies support only services relevant for the query establishing relations between terms (low-cost flights, travel agencies, booking, etc.) Limited expressivity (stringent computational requirements) (Pre-processing time) Meaning negotiation and explanation Foundational Ontologies establish consensus about meaning of a new term within a community Explain meaning of a term to somebody new to community Higher expressivity and rich axiomatization needed to exclude ambiguities Only needs to be undertaken once, before cooperation process starts. 18
The Conceptual Bridge for Meaning Negotiation Meaning Negotiation aims at finding an agreement on the meaning of the (verbal) expressions that are used in explicit communication among hybrid agents (P2P, booking of flights, buy/sell) Open multi-agent systems involve different conceptualizations of the external environment Requirements for building a bridge between multiple conceptualizations: Solid engineering methodologies and reliable tools Ontologies have the aim of formally characterizing the intended meaning of terms, favoring communication and machineunderstandable information sharing within agents. 19
Domain Ontologies Physical objects Topological, morphological, cinematic, and functional features of physical objects Information and information processing Ontology of semiotic relations Ontological nature of data, documents, files, pointers Ontological foundations of subject classification Social interaction Ontology of planning and control Ontological aspects of trust, delegation, autonomy, cooperation Ontology of organizations Legal and financial entities Ontology of norms and contracts Ontology of transactions Ontology of speech acts and dialogue relations 20
Ontology-driven Information Systems Ontology-driven conceptual modeling Unified methodology for ontological analysis Criteria for evaluating, comparing, and re-engineering ontologies Impact of ontological analysis on UML modeling Ontology-driven information access Use of lexical resources for multilingual information access Ontology-driven information integration Unified methodology for the integration of terminologies and thesauri 21
Examples of ontological distinctions
General examples Examples from human common sense They are useful to understand the basic characteristics of ontological analysis and insight Generic distinctions, such as substance/object, quality/quality value, feature/object, are important for building Foundational Ontologies 23
Substances and Objects Situation: Starting from the scattered snow (amount of matter) covering his courtyard, John can make-up a snowman (object), adding a hat, a carrot, two deadwoods, etc. Ontological description: what distinguishes an object from an amount of matter is that the former is an essential whole, namely it has a unity criterion, while the latter is not. 24
Qualities and Quality regions The rose and the hair band have the same color: different color qualities inhere to the two objects they are located in the same quality region Therefore, the same color attribute (red) is ascribed to the two objects 25
Features of objects Features are parasitic entities, that exist insofar their host exists. Features may be relevant parts of their host, like a bump in a road, or dependent places, such as a hole in a piece of cheese, the underneath of a table, or the shadow of a tree (which are not parts of their hosts). 26
Specific examples Lightweight ontology Foundational Ontology Domain Ontology All examples have been implemented using Protégé, software tool for conceptual modeling and ontology creation developed by Stanford University (USA). 27
Computational ontologies
Ontology Languages Ontology languages allow users to write explicit formal specifications of conceptualizations related to a given domain. RDF, RDF(S), DAML-ONT, OIL, DAML+OIL OWL 1.0 (W3C standard) Core requirements: Well-defined syntax Well-defined semantics Efficient reasoning support Sufficient expressive power 29
Syntax Well-defined syntax is important for enabling machineprocessing of information RDF syntax (XML-based) is hard to be considered as user-friendly! Instead of directly writing the code, users exploits ontology development tools, such as Protégé, Swoop, Top Braid Composer, 30
Semantics Well-defined (formal) semantics describes rigorously the meaning of knowledge It allows for precise reasoning. For ontological knowledge, we can reason about Class membership: if x is an instance of 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 equivalent to class C, then A is equivalent to C, too. Consistency: given x instance of class A and suppose that: A is a subclass of B C A is subclass of D B and D are disjoint then we have an inconsistency the ontology must be fixed Classification (or subsumption reasoning): if we have declared that certain property/value pairs are sufficient condition for membership of a class A, and if x satisfies these conditions, we can conclude that x must be instance of A. 31
Efficient reasoning support Inferences like the one presented before can be made mechanically, instead of manually. Reasoning support is important for Check the consistency of an ontology Check for unintended relationships between classes Automatically classify instances in classes Automatic reasoning help for Designing large ontologies (eventually involving multiple authors) Integrating and sharing different ontologies 32
Ontology Web Language (OWL) Formal semantics and reasoning support are provided by mapping ontology language to logical formalisms and using automated reasoners that already exist for those formalisms. OWL is (partially we ll see why) mapped on a description logic, and makes use of existing reasoners such as FaCT and RACER. Description logics are a decidable fragment of First Order Logics. REMINDER: Logics are decidable if computations/algorithms based on the logic will terminate in a finite time. 33
OWL sub-languages OWL-LITE Syntactically simplest used for class hierarchy and simple constraints OWL-DL Based on DL used when automated reasoning is needed OWL-Full Most expressive used when expressivity is more important than reasoning 34
Appendix: OWL primitives
Class In the following we introduce some examples of OWL core components; most are taken from the ontology pizza.owl (we ll see how this ontology looks like in the Protégé-OWL platform). <owl:class rdf:id="anchoviestopping"> <rdfs:subclassof rdf:resource="#fishtopping"/> <owl:disjointwith rdf:resource="#prawnstopping"/> <owl:disjointwith rdf:resource="#mixedseafoodtopping"/> </owl:class> OWL extends RDF: it adopts RDF meaning of classes and properties, extending the expressivity with its own primitives owl:thing the most general class (it subsumes every class); owl:nothing the empty class (every class subsumes it) 36
Object Property Object property relates classes (objects) to other classes (objects): <owl:objectproperty rdf:about="#hastopping"> <rdfs:subpropertyof> <owl:objectproperty rdf:about="#hasingredient"/> </rdfs:subpropertyof> <rdfs:domain rdf:resource="#pizza"/> <rdfs:range rdf:resource="#pizzatopping"/> <owl:inverseof> <owl:objectproperty rdf:about="istoppingof"/> </owl:inverseof> </owl:objectproperty> InverseOf interchange domain with range 37
Datatype Property Datatype property relates classes (objects) to datatype values: <owl:dataproperty rdf:id="#table_number"> <rdfs:range rdf:resource= http://www.w3.org/2001/xmlschema#nonnegativeinteger"/ > </owl:dataproperty> OWL does not provide predefined datatypes: it uses XML Schema datatypes Language layering: XML XML(S) RDF RDF(S) OWL 38
Do you want more?
Useful Links http://www.loa-cnr.it http://www.fois.org/ http://protege.stanford.edu http://www.cyc.com/ http://ontology.buffalo.edu/ http://www.jfsowa.com/ontology/guided.htm 40
References G. Antoniou and F. van Harmelen, Web Ontology Language, Handbook on Ontologies, Springer, 2004. M. Horridge, H. Knublauch, A. Rector, R. Stevens, C. Wroe, A practical guide to building OWL ontologies using The Protégé-OWL plug-in and CO-ODE Tools Edition 1.0. Available at: http://protege.stanford.edu/ http://protege.stanford.edu/plugins/owl/publicatio ns/2004-07-06-owl-tutorial.ppt http://www.w3.org/tr/owl-features/ http://www.sts.tu-harburg.de/r.f.moeller/racer/ http://www.mindswap.org/2003/pellet/ http://www.cs.man.ac.uk/ horrocks/fact/ 41