just enough ontology tutorial by paola di maio wims, 25-27 may 2011, norway

87
JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

Upload: lynne-lang

Post on 12-Jan-2016

217 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JUST ENOUGH ONTOLOGY

TUTORIAL 

by Paola Di Maio

WIMS, 25-27 MAY 2011, NORWAY

Page 2: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

OVERALL OUTLINE 

• Introductions, Tutorial overview 15 minutes• Problem space, common challenges 5• A few words about innovation, knowledge, open

innovation • Why ontology, and overview of main concepts• Ontology engineering (development)• Just Enough Ontology

Page 3: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

Scope of this tutorial

WHAT IT IS:  AN AGILE ,  PRAGMATIC, SOCIO-TECHNICAL APPROACH TO  ONTOLOGY DEVELOPMENT /A REFLECTIVE, EVOLUTIONARY LEARNING FRAMEWORK 

WHAT IT IS NOT: NOT INTENDED TO BE A COMPLETE APPROACH, IT DOES NOT INTEND TO ADDRESS EXAUSTIVELY NOR SATISFY IN DETAIL EVERY ONTOLOGY ENGINEERING REQUIREMENT 

Page 4: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

INTRODUCTIONS!

Say a few words about yourself, where you come from,what brings you here, your interest, issues and expectations  for this tutorial, if any

Then we can take a look at the planned agenda for today, and make adjustments if necessary

Page 5: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

 KNOWLEDGE CHALLENGES

INFORMATION OVERLOADVERY FAST KNOWLEDGE EXCHANGESVERY FAST DEVELOPMENT CYCLESCANT KEEP UP WITH PROGRESS ALL AROUNDCONVERGENCE OF MANY DISCIPLINESDIFFICULT TO STAY ON TOP OF EVERYTHINGTOO MUCH KNOWLEDGE TO GRASP/REASON WITHVERY RAPID CHANGES, SHORT ITERATIONS MAKE PROJECT PLANNING DIFFCULT

Page 6: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

WE DESIGN INFORMATION SYSTEMS TO COPE

However, an information system needs a  conceptual frame, so../

Page 7: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

WE DEVISE ONTOLOGIES TO SUPPORT SYSTEM ARCHITECTURES

However, there are many ways to go about developing an ontology, which one to choose?

Do we hire an ontologist?

What level of resources should be devoted to the ontology development? 

How long will it take?

How do we evaluate the ontology?etc... etc..

Page 8: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

all partly right but on the whole  all wrong

Page 9: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

 A FEW WORDS ABOUT KNOWLEDGE,

INNOVATION, OPEN INNOVATION

Page 10: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

INNOVATION

TO GENERATE NOVELTY (INCORPORATE THE NEW)

INCREMENTAL/EMERGENT

SUBSTANTIAL

RADICAL/REVOLUTIONARY

Page 11: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

THE ROLE OF KNOWLEDGE IN INNOVATION"Innovation . . . is generally understood as the successful introduction of a new thing or method . . . Innovation is the embodiment, combination, or synthesis of knowledge in original, relevant, valued new products, processes, or services."   Luecke and Katz (2003)

Page 12: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

OPEN INNOVATION

THE ABILITY OF AN ORGANISATION TO DEVELOP, ACCESS, INTEGRATE, DEPLOY KNOWLEDGE

Open Innovation,  A New Paradigm for Understanding Industrial Innovation, Henry Chesbrough http://www.openinnovation.net/Book/NewParadigm/Chapters/01.pdf

Page 13: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

IBM Global CEO Study 2006

Page 14: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

PEOPLE AND THE OPEN WORLD

PEOPLE: BEHAVIOURAL, SOCIO-TECHNICAL DIMENSIONS

OPEN WORLD (VS CLOSED WORLD)DIFFERENT ASSUMPTIONSCWA holds that any statement that is not known to be true is false.OWA  no single agent or observer has complete knowledge, any statement that is not known to be true, is not necessarily falsehttp://en.wikipedia.org/wiki/Open_world_assumption

Page 15: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ONTOLOGY

Page 16: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

WHAT IS AN ONTOLOGY ANYWAY(...... oh no not again...)

Ontologies are conceptual and semantic representations widely used, in different forms, to capture and express such models. 

.... SEE HANDOUT FOR A LIST OF DEFINITIONS AND ADD YOURS ......

Page 17: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ontology/definitionThe etymological source of the term“ontology” — ont — comes fromthe Greek verb “einai” (to be)from which the Latin word followed— “ontologia” (ont +ogia), translated as “the study of existence” [15]. Ontology is writtenwith an uppercase initial when referring to the top-level “discipline” and with a lowercasein all other occurrences [14].  or the study of what we know exists>

MORE IN HANDOUTS

Page 18: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

WHY ONTOLOGIES

Information centric systems  are designed to leverage knowledge expressed via natural language symbols and meanings (semiotics and semantics)  need to be captured and represented adequately  for these systems to function.

Page 19: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ONTOLOGY AS THE ULTIMATE KNOWLEDGE SPECIFICATION

ontology = detailed, arguably complete, but always comprehensive knowledge schema

many definitions, views , see separate handout

HOWEVER DEVELOPING AN ONTOLOGY IS NON TRIVIAL

Page 20: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

THE PROBLEMA

Many methodologiesSkills are scarcevery complex projectpeople with different views/prioritiescognitive requirements are highorganisation/coordination is difficultno single resource to helps us devise an ontology (more...)

Page 21: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

 ONTOLOGY ENGINEERING IS A SPECIALISED SKILL

AN ONTOLOGY COULD BE THE ANSWER TO MANY INFORMATION SYSTEMS CHALLENGES, BUT PARADOXICALLY, ONTOLOGY DEVELOPMENT (AKA ONTOLOGY ENGINEERING) TENDS TO BECOME A LONG TERM PROJECT IN ITSELF, IE, REQUIRES DEDICATED ALLOCATION OF RESOURCES, lots of fluff in there as well

THEREFORE IN THIS TUTORIAL WE SUGGEST A 'JUST ENOUGH ONTOLOGY' APPROACHTO TRY TO GET TO OUR SOLUTION WITHOUT GETTING LOST

Page 22: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

AN ONTOLOGY CONSISTS OF...

An ontology may take a variety of forms, but it will necessarily include a vocabulary of terms and some specification of their meaning, such as definitions and an indication of how concepts are interrelated, which collectively impose a structure on the domain and constrain the possible interpretations of terms 

Ushold et al

Page 23: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

PROLIFERATION

Many ontology engineering (OE) methods, artifacts, tools and techniques do not make ontology has not become any easier. The choice of an appropriate methodology for any project may require a systematic  evaluation of existing approaches, and this can become extremely resource intensive and time consuming. A default option is to follow no methodology at all (the 'who needs a methodology?' attitude) often preferred by developers who go straight into coding, and may confuse 'on the fly' schema creation with fully fledged ontology development.

Page 24: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JUST ENOUGH APPROACHES...

'SOFT' AND 'AGILE' APPROACH DEVELOPED BY ED YOURDON IN THE SEVENTIES

WHEN DATABASES BECAME THE KEY TO INFORMATION SYSTEMS, STRUCTURING DATA BECAME THE NEXT BIG THING, BUT TOO MANY METHODOLOGIES, AND TOO RIGID APPROACHES, MADE THE TASK OFTEN MORE BURDENSOME THAN THE CHALLENGES THEY WERE TRYING TO SOLVE

Page 25: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

some useful PRINCIPLES

AGILITY

ADAPTIVITY

EVOLUTIONARY APPROACH ----------------------JUST  IN TIME

JUST ENOUGH

Page 26: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

IS AN ONTOLOGY A VOCABULARY?

http://infogrid.org/wiki/Reference/PidcockArticleA controlled vocabulary is a list of terms that have been enumerated explicitlyA thesaurus is a networked collection of controlled vocabulary terms. 

People use the word ontology to mean different things, e.g. glossaries & data dictionaries, thesauri & taxonomies, schemas & data models, and formal ontologies & inference. A formal ontology is a controlled vocabulary expressed in an ontology representation language, ... a grammar for using vocabulary terms to express something meaningful within a specified domain of interest. The grammar contains formal constraints (e.g., specifies what it means to be a well-formed statement, assertion, query, etc.) on how terms in the ontology’s controlled vocabulary can be used together.

Page 27: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

 

Page 28: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

applied minds

Page 29: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

The notion of boundary

    WHAT CONSTRAINS A SYSTEM

   (lexical, conceptual, etc)

Page 30: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

MANY TYPES OF ONTOLOGIES

 Upper ontology - toplevel categories and classesDomain ontology Aspect ontology — aspect of a domain Formal ontology — when the ontology is fully formalizedInformal ontology — any structured collection not constrained by a formalizationApplied ontology — wherethe ontology is developed insupport of software artifactsTask ontology — when it isdevised in support of a specifictask or function, as opposed to a whole domain 

Page 31: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

MORE TYPES OF ONTOLOGIES

1. Content ontologies for reusing knowledge2. Communication ontologies for sharing knowledge3. Indexing ontologies for case retrieval4. Meta-ontologies (knowledge representation knowledge)

Van Heijst et al. identifies orthogonal dimensions todistinguish different types ofontologies: (1) terminologicalontologies, information technologies, and knowledge modelingontologies; and (2) representation, generic, domain,

Page 32: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

TAXONOMY OF THERMAL COMPONENT 

Page 33: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

TAXONOMY OF PHYSICAL MECHANISM

Page 34: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

CONCEPTUAL SCHEMA OF SYSTEM

Page 35: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ONTOLOGY DEVELOPMENT LIFECYCLE

Page 36: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

MANY METHODSDEF5TOVEDODDLECLEPE/AFMCyc methodMike Uschold and Martin King’s methodMichael Grüninger and Mark Fox’s methodKACTUSMETHONTOLOGYSENSUSOn-To-KnowledgeOnto CleanDILIGENT,HCOME,  OTK methodology,Ontology Development 101CO4,KASquare,DOGMA (AKEM)SEKT,OnTo Knowledge(OTK)OntoCleanBORO

Page 37: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

METHODOLOGICAL DIFFERENCES

THE PHASES TEND TO BE THE SAMESOME DETAIL CHANGESTHE MAIN DIFFERENCE IS THE SCHEDULING OFDEVELOPMENT ACTIVITIESSOMETIMES DIFFERENT EMPHASES

Page 38: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ONTOLOGY PROJECTS CAN BE HIGH RISK OF FAILURE

BECAUSE OF • VERY HIGH COMPLEXITY• MANY UNCERTAINTIES• OD/E SKILLS ARE SCARCE

RISK OF OVERENGINEER/GET LOSTOR UNDERENGINEER/OVERSIMPLIFY

WE ARE LOOKING FOR THE MIDDLE WAY

Page 39: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JUST ENOUGH OD/E RATIONALE

CAPTURES COMMON DENOMINATOR TO MOST METHODOLOGIES

MAKES ACTIVITIES AGILE/ITERATIVE

SIMPLIFIES THE PROCESS BY PROVIDING A SCHEMA

KEEPS ALL OPTIONS OPEN

DESIGNED TO CAPTURE EMERGENCE

Page 40: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

PRINCIPLES: STRUCTURING AND ABSTRACTION

In data modeling, abstraction is what allows to identify and group information assets based on generic common characteristics that exist independently from their time/space representation. knowledge modeling Object Oriented Design (OOD), Unified Modeling Language (UML), Integrated Development Framework (IDEF)

Page 41: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

SINGLE ABSTRACTION

Structured methods rely on the notion of 'single abstraction' mechanism, which consists of extracting a top level view of different aspects of the system, forming the basis for functional decomposition', the technique that drills into top level functions, and breaks them down into smaller functions, while preserving the representation of other functional aspects of the systems such as inputs, outputs, controls and other mechanisms. 

Page 42: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

DIAGRAMMATIC REPRESENTATION

Diagrammatic methods  such as UML, for example, can be  used as a form of ontological notation (S. Cranefield), although it is sometimes  argued that they may not have the 'expressivity' required  to represent all of the essential ontological formalism, such as axioms. Topic Maps are an ISO information standard that enables multiple, concurrent views of sets of information objects which can be used  to represent ontological knowledge, Concept maps and mind maps are also used to aid conceptual visualizations, although they are not  a standardized. 

Page 43: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

PROBLEMS WITH STRUCTURED METHODS

HEAVY, NON AGILE

PERCEIVED AS SUITABLE ONLY FOR WATERFALL DEVELOPMENT STYLE

Page 44: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

DESIGN PATTERNShttp://en.wikipedia.org/wiki/Design_pattern_(computer_science)

WAY OF REPRESENTING A STRUCTURE FOR REUSEC ALEXSANDER, ET ALAlgorithm  how to exploit application characteristic on a computing platform.Computational  addressing concerns related to key computation identification.Execution patterns that address concerns related to supporting application execution,Implementation strategy patterns addressing concerns related to implementing source codeStructural design patterns addressing concerns related to high-level structures of applications being developed.

Page 45: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

ONTOLOGY PATTERNS (Blomqvist)

Application Patterns - Purpose, scope, usage and context of the implemented ontology or ontologies, including interfaces and relations to other systems.Architecture Patterns - A description of how to combine or arrange implemented Design Patterns in order to fulfill the overall goal of the ontology.Design/Semantic Patterns - A small collection of Semantic Patterns that together create a common and generic construct for ontology development.Syntactic Patterns - Language specific ways to arrange representation symbols, to create a certain concept, relation or axiom. (end sidebar) 

Page 46: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JOE MAIN CONTRIBUTIONS

1) a strong emphasis on stakeholder analysis and management

2) implementation independence

Page 47: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

STAKEHOLDERS

A diverse stakeholder basis  necessary to  a balanced mix of views and sustainability of ontologies, especially their use and long term maintenance. OE requires depth and breadth of understanding, knowledge and skills in a variety of fields.  It is becoming increasingly important to broaden the stakeholder base and to make this process accessible to as many participants as possible, but not at the expense of validity and ‘ontological rigour’, although even validity and rigor depend on where certain boundaries are set. takeholders bring onto the development table a much needed socio-technical perspective, of which people and environments are important elements.

Page 48: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

IMPLEMENTATION INDEPENDENCE

A variety of stakeholders with different skill sets (not necessarily skilled in the implementation language of choice) and used as part of sometimes heterogeneous architectures,  separation between ontology design and itimplementation.  Implementation independence is a well established principle in systems science, and it constitutes one of the core features of systems architectures. In the early days of computing it was advocated by Childs 

Page 49: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

IMPLEMENTATION INDEPENDENCE 2

Codd  and Dates relational model as 'data independence'

Model Driven Architectures 'platform independence'

In Ontology, Tom Gruber advocated ‘minimal encoding bias’

Page 50: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

1. Clarity2. Coherence 3. Extendibility4. Minimal encoding bias5. Minimal ontological commitment 

GRUBER'S 5 PRINCIPLES

Page 51: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JEO, 13 STEPS (mostly iterative)1 Identify stakeholders, outline stakeholder profile2 Define the purpose of the ontology (emphasis on representation/indexing, problem solving/reasoning)3 Outline requirements [iterative]4 Identify and survey existing knowledge sources and existing ontologies,elicit existing knowledgeAssess why the existing knowledge resources do not meet the intended user requirements,update the requirements with the output of activities above [iterative5 Scoping ontology (defining the boundaries andlevel of granularity, according to goal and stakeholders requirements) Update the requirements  [iterative]6 Devise and implement quality assurance planAdd quality parameters to the requirements  [iterative]

7 Define the field of competence to identify the knowledge boundaries (competence questions)Match the field of competence with the knowledge sources8 Define the ontology artifacts:VocabularyIdentity concepts/entities/classesrelations, axiomsRefine and map vocabularies to artifact9 Transfer concepts to ontology language representation: Select knowledge representation  formalisms and annotation depending on stakeholder requirements, scope and goal10 Deploy/systems integration (modular, incremental)11 Testing Evaluation quality monitoring competence assessment12 Publishing13Maintenance Reuse

Page 52: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

1. IDENTIFY STAKEHOLDERSPotentially, anyone actively involved in the ontology development and intended use, and anyone whose interests may be affected by the development of such an ontology.Stakeholder management'  to assess goals, motives and commitments, and to create and sustain the collaborative momentum that can fuel an ontology development project itself.For example: assessing  their ‘readiness’ to participate in the project,identifying  the barriers that prevent them from participating and making commitments to the ontology, determining the congnitive or organizational requirements, including existing and potential conflicts with existing conceptual models in use.

Page 53: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

2. DEFINE GOAL/PURPOSE

GIVEN THE MULTIPLICITY OF THINGS THAT ARE UNDERSTOOD UNDER 'ONTOLOGY' (SEE THE EXAMPLE BEFORE, IT IS IMPORTANT TO ESTABLISH WHAT IS THE INTENDED USAGE OF THE ONTOLOGY

examples: support a process execution within a  systemImprove the efficiency of reasoningConsolidate and harmonize existing data/informationProvide an abstract, more schematized viewCreate a consensual, unified view that can serve as synthesys of different viewsProvide a formal specificationSupport integration of data, applications, and systems to help minimize design and planning errors caused by lack of domain knowledge

Page 54: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

3.OUTLINE REQUIREMENTSSHALL/SHOULD....FOLLOW USUAL RASUPPORT THE STATED GOALS/PURPOSEDESIRABLES:• It must  declare explicitly what high level knowledge (upper

level ontology) it references,• Must declare explicitly what kind of reasoning/inference

supports/it is based on.• It must be accessible to all the agents/agencies (this means

shared, viewable, understandable)• It must be ‘acceptable’ to all the agents/agencies from the

different perspectives, in terms of culture, language, conformance to policy and protocols

• It must be ‘usable’ in terms of compatibility with local information systems used by agents/agencies

Page 55: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

MOST ONTOLOGIES WILL BE USED FOR:Consistency checking ( properties and value restrictions)Auto Completion of information partially provided by usersInteroperability support (shared conceptualization)Support validation and verification testing of data (and schemas)Configuration support — class terms may be defined so that they contain descriptions of what kinds of parts may be in a system.

Page 56: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

4. KNOWLEDGE SOURCESDECIDE AND MAKE EXPLICIT WHERE DOES THE K COME FROM

Stakeholders (who have a vested interest in the business, application, or component being specified).Subject matter experts (e.g., domain experts, industry analysts, consultants).Reusable requirements and requirements specifications (see Requirements Reuse)Documentation (e.g., business strategy documents, business vision statement, documentation of relevant legacy systems, workflow procedures, vendor documentation, marketing studies, change/enhancement requests from the users and technical service representatives, industry analyst reports, etc.).Human Interface Prototypes.Legacy, competing, and similar applications.Application and enterprise databases

Page 57: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

5. SCOPE THE ONTOLOGY: LEVELS OF REALITY

Page 58: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY
Page 59: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

LATTICE OF TOP LEVEL CATEGORIESJ SOWA http://www.jfsowa.com/ontology/toplevel.htm

Page 60: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

MEREOLOGY

Theory of whole and parts

Page 61: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

Component level: Out of which concrete artefacts (device components) does the system that is to be designed exist, and how are they interconnected (system structure or layout)? Process level: How is the behaviour of the system components realized in terms ofphysical mechanisms? Mathematical level: How is the physical behaviour formally specified in terms of equations,such that system analysis and simulation can be carried out by computer

Page 62: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

6. QUALITY MODELQuality Models contain patterns of qualitative and quantitative measurements of various aspects, and quality cannot be easily assessed with straight ‘testing’, ad quality often results from a combination of factors, of which ‘correctness’  and ‘accuracy’ are merely part of.  A ‘ quality model’ is developed upfront, and its dimensions and metrics are used as target parameters throughout the development, evaluation and testing of an ontology. The best measure for quality is ‘fit for purposeness’, also know as the 'good enough' principle, however measuring standard parameters can help develop a quality strategy which is essential especially when working in large organizations. In a narrow sense, the quality of an ontology is  measured across two dimensions:  accuracy and its comprehensiveness, corresponding roughly to precision and recall in search technology, but a whole range of parameters can be used to assess the quality, as shown in the indicative table below.

Page 63: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY
Page 64: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

7 FIELD OF COMPETENCE

DOMAIN OF COMPETENCE = KNOWLEDGE FIELDSET OF QUESTIONS TO VERIFY WHETHER AN ONTOLOGY CAN ANSWERS THE QUESTIONSGET THE QUESTIONS FROM THE STAKEHOLDERS Different tests can be set up to verify the validity and quality of each part of the ontology, and each process within the ontology development, and carried out correspondingly at each step. 

Page 65: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8. ARTEFACTS

1. vocabularies2. concepts3. relations4. axioms/ruls

Page 66: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/1 VOCABULARIES

STUDENTdefinition: person who studies |class, subclass of_x |relation: part_of, parent_of |allowed value: male, female|

Page 67: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/2 CONCEPTSConcepts are fundamental to our ability to think, express, represent and communicate, however, defining unambiguously and with certainty what constitutes a concept, is rather tricky, and pushes IT practitioners toward the realm of philosophy. But that’s a challenge of ontology engineering.  Concepts can correspond to things, but also to ‘fuzzy clouds’ of ideas and notions identified by words and related to a certain thing or subject. And even when referring to tangible things, concepts can be abstract, volatile. Even simple ‘things’ are made of parts, relations, dependencies, temporal and spatial circumstances. When trying to put the finger exactly reality, all sorts of questions inevitably come up. The nearest technique that can be compared to conceptual modeling, is entity modeling, or class modelling. Concepts can be broadly divided into cognitive artifacts that support categorization and communication, [20] and are necessary to support human and artificial thinking and reasoning. The purpose of ontologies is to make them explicit and represent them so that they serve a variety of purposes, namely the intended goals. Conceptual categories and thoughts are closely related to language. A concept model can be used to complement and extend a functional data model.

Page 68: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/2.1 RELATIONS

• Arity: Typically binary relationships are of most interest, but relationships can be of arbitrary arity, i.e., we could have 3 or more concepts participating in a relationship.These constraints are characterized in one of the following ways: 1-1, many-1, 1-many, or many-many. A more generalized way of representing these cardinality constraints is using a pair of numbers that specify the minimum and maximum number of times an instance of a concept can participate in a relationship. This is avery useful technique for n-ary relationships and also captures partial participation of concepts in relationships. 1-1 and many-1 relationships are functions which can be exploited in various ways.• Direct v/s Transitive Relationships: • Crisp vs. Fuzzy:

Page 69: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/2.2 RELATIONScausal temporalspatial relationsemantic relationlogical relation

In a language context, relations are called ‘lexical relations’, and are known as ‘taxonomic relations’, such as synonymy, where two different terms point to different concepts, homonomy, where the same word points to two different things or concepts, antonymy, where two terms indicate two opposites etc.

Page 70: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/2.3 example of lexical relationsfrom OBO (Stanford) 

Page 71: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/4 AXIOMS

 WHAT IS ALWAYS TRUEGENERALLY EXPRESSED AS RULESUSED TO CHECK CONSISTENCY SUPPORT OF REASONING

Page 72: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

8/4.2 AXIOMS AS RULES

Axioms translate into constraints, which in turn can be considered as the logical boundaries of an ontology. They can be transformed and mapped easily directly into rules.  If an axiom maps to a rule, then consisting parts of an axiom map to the consisting parts of a rule. The mapping follows:

• ontology axiom → rule• axiom statement → rule clause• statement concept → entity in a rule clause• statement relationship → relationship in a rule clause

Page 73: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

9 FORMALISM/IMPLEMENTATION

A FORMALISM CAN BE DEFINED AS A  'LANGUAGE/NOTATION'RULE 4 OF GRUBER (SEE GRUBER'S 5 RULES)DO NOT OVER COMMIT TOO EARLY

• DESCRIPTION LOGIC• XML/RDF/OWL• COMMON LOGIC• OTHERS (...)

Page 74: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

9/1 DESCRIPTION LOGIC

Page 75: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

9/2.2 TBOX AND ABOX

Knowledge representation system based on DLs consists of two components - TBox and ABox. 

The TBox describes terminology, i.e., the ontology in the form of concepts and roles definitions, while the ABox contains assertions about individuals using the terms from the ontology. 

Concepts describe sets of individuals, roles describe relations between individuals

Page 76: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY
Page 77: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

10. PUBLISHING

MAKING DISCOVERABLE,AVAILABLE, ACCESSIBLELEVEL OF PERMISSIONING (COPYRIGHT, INTELLECTUAL PROPERTY)

SENSITIVITYFORMATTINGINDEXING TAGGINGLINKING!  

Page 78: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

10.1 OWL: ontology web language

W3C SPECIFICATIONhttp://www.w3.org/TR/owl-guide/ LOOK AT THE SPECOWL Web Ontology LanguageNew Version Available: OWL 2 (Document Status Update, 12 November 2009)The OWL WG has produced a W3C Recommendation for a new version of OWL which adds features to this 2004 version, while remaining compatible. Please see OWL 2 Document Overview

Page 79: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

THEN WATCH DEMO WEBCAST

SEE THE WEBCASThttp://knoodl.com/ui/site/webcast/intro.jsp

Page 80: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

11. INTEGRATION

DEPLOY THE ONTOLOGY AS PART OF A SYSTEM

INTERFACES (SYSTEMS, USERS ETC)

Page 81: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

12. EVALUATION/TESTING

Consistency, integrity, validation, redundancy testing, as well as usability testing can be adapted to test an ontology. In order of priority, an ontology should be tested for -accuracy – ability to support correct inference – in relation to the expected ‘range of competence’ -its ability to identify and represent correctly linguistic intensionality and extensionality, for example, where the first refers to the ‘aboutness’ of an expression, and the extension is  the class of objects that an expression refers to.http://en.wikipedia.org/wiki/Extension_(semantics)http://en.wikipedia.org/wiki/Intension

Page 82: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

13. MAINTENANCE

REVISE PERIODICALLYUPDATESET UP FEEDBACK LOOPCAPTURE ITERATIVE CHANGESSTART FROM 1, REPEAT AS MANY TIMES AS NECESSARY

Page 83: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

Coverage/ScopeIs the vocabulary capable of representing all of the concepts used in the chapter?Does the vocabulary have the terms necessary to represent the full range of issues?Does the vocabulary encompass the terminology used to describe the various procedures?Does the vocabulary use terms that are commonly used by SE?

SpecificityIs the vocabulary specific enough to accurately represent the many aspects of SE reality?Does the vocabulary capture information in sufficient detail?

StructureAre the vocabulary hierarchies logical and complete?Are the meanings of terms clearly defined?Does the vocabulary contain redundant terms?Are there explicit rules for combining terms, or for combining terms and qualifiers?Does the vocabulary allow for multiple classification of terms, that is, can terms appear in more than one hierarchy?

UseabilityIs the vocabulary mapped to other vocabularies used in the practice?Does the vocabulary meet the needs of a range of end users?Does the user interface facilitate optimal use of the vocabulary with minimal training?

Page 84: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

RECOMMENDED TOOLS TO TRY

http://knoodl.com

SEE THE WEBCASThttp://knoodl.com/ui/site/webcast/intro.jsp

Referata

PROTEGE

Page 85: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

CONCLUSION/QUESTIONS

this tutorial introduces an agile approach to ontology developement in relation 

questions?

Page 86: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

JUST ENOUGH ONTOLOGY

WWW. JUSTENOUGHONTOLOGY.ORG

OPEN FOR COLLABORATIVE DEVELOPMENT

ADD YOUR OWN ITERATION, RESOURCE, ILLUSTRATION, CASE STUDYETC...

Page 87: JUST ENOUGH ONTOLOGY TUTORIAL by Paola Di Maio WIMS, 25-27 MAY 2011, NORWAY

EXERCISE/DISCUSSION

IF THERE IS TIME, FROM A GROUP, ASSIGN ROLES (STAKEHOLDERS)CARRY OUT STEPS 1/13 WHERE POSSIBLEDISCUSS