EDON: A Method for Building an Ontology as Software Artefact

Posgrado

EDONA METHOD FOR BUILDING AN ONTOLOGY

AS SOFTWARE ARTEFACT

Emiliano Reynares, María Laura Caliusco, María Rosa Galli

Doctorado en Ingeniería – Mención Sistemas de Información

CIDISI UTN FRSF

Posgrado

AGENDA

1. The problem

2. Overview of EDON

3. EDON Method

4. Experiences from a study case

5. Discussion and future research directions

Posgrado

THE PROBLEM

Recognizing the context

• Environmental business dynamics

• “Embededness” of business rules into software applications

Posgrado

THE PROBLEM

A proposal

Encapsulate business rules in a specific system component

RULES BASES, but…

• using different attributes to represent the same thing

• defining rules based on incompatible notions of the same concept

could affect the rule interaction [Zacharias, 2008]

Posgrado

THE PROBLEM

What about ontologies?

As artefacts intended to explicitly represent the data semantics, could encapsulate business rules while avoiding concept misunderstanding

BUT…

• Ontology building process must be considered in the context of the software development project

• Conceptual dynamics issue needs to be tackled [Hepp, 2007]

• A descriptive path enable to quickly and smoothly evolve through further extended versions of the ontology

Posgrado

EDON Evolutive Development of ONtologies

Pursues three goals

1. To develop from scratch an ontology intended to encapsulate the declarative specification of business rules, supporting accurate and well-defined requirements of a software system

2. To enable the intertwining of ontology and software development process, facilitating a quick integration of both artefacts

3. To allow the ontology readily reflect the changes in quickly evolving domains by providing a phase model to evolve through successive versions of the ontology

Posgrado

EDON

Collaborately performed by

1. Domain ExpertsDEs interact daily with the problem domain, so they have the knowledge to be modelled

2. Knowledge EngineersKEs have the know-how for representing the reality in a way that supports the requirements of a software application

Modelling activities are performed at a high level of abstraction to facilitate the communication between these profiles

Short development iterations allow to readily reflect conceptual dynamics and changes in requirements, and both profiles can take advantage of what was learned during the building of earlier versions of the ontology

Posgrado

EDON

In the context of software development process

Posgrado

EDON

The process

Posgrado

EDON

Requirements Selection Process

Posgrado

EDON

Requirements Selection Process in a study case

• Functional requirements selected considering two issues: 1. Ontology should provide ontology-based reasoning over business rules 2. Human users of the software application should not interact directly with the

underlying ontology but through the software application.

• “Project”, “Researcher”, and “Funding agency” were considered the core entities

• A storyboard exposing a functional requirement selected for further development:“The software system should allow a researcher to upload a research project pending approval. During this upload, the software system must validate the input data against the rules defined by the agency intended to funding the project”

Posgrado

EDON

Ontology Development Process

Posgrado

EDON

Ontology Specification Activity in a study case

Competency Questions [Gruninger et al, 1995]

CQ1 What are the requirements a research project must satisfy to be accepted for a funding agency?

CQ2 What are the requirements the manager of a research project must satisfy?

CQ3 What are the requirements a member of a research project must satisfy?

CQ4 What are the documents evaluated for a funding agency?

Entities

Research Project Researcher Funding Agency

Project Member Project Manager Evaluation Documents

Posgrado

EDON

Ontology Conceptualization Activity in a study case• Representation of the knowledge associated to the domain entities by using the Lexicon

Extended Language (LEL) [Leite et al, 1993]

• Representation classified in four categories: “object”, “subject”, “verb”, and “state”

• Term described by “notion” and “behavioral response”

The Project Manager glossary entry

Notion

It is a Researcher at Category I or II, working in an Execution Unit and managing, at most, two Research Projects. Each Research Project requires, at least, twenty hours-month dedication.Behavioural response

To develop research projects; To raise funds for the project execution; To generate reports on project status

Posgrado

EDON

Ontology Formalization Activity in a study case

• Generation of a base ontology by performing a systematic process [Breitman et al, 2004]

• It gives an ontological structure to the terminology described in the conceptualization activity

• The structure is independent of any ontology implementation language but can be mapped into most of them

Concepts Relations Axioms

Research Project

HasCodeHasTitleHasThemeHasManagerHasManagerDedicationHasReport

The manager is a researcher at category I or IIThe manager must dedicate, at least, twenty hours-month at each managed project.

Posgrado

EDON

Ontology Refinement Activity in a study case

• Extending the base ontology by focusing on the formulation of axioms

• Knowledge obtained from the information sources identified

• DEs collaborate with the aim of producing the tightest possible model of reality

• KEs are focused on supporting a set of functional requirements

Axioms

If a project report has a 30 day delay, the research project is cancelled

If a project report has a 30 day delay, the project manager is suspended

If a project report is rejected, the research project is cancelled

Posgrado

EDON

Ontology Implementation Activity in a study case

• Encoding of the ontology in an implementation language

• Evaluation increases along development stages. Three dimensions must be evaluated [Burton-Jones et al, 2005] [Gangemi et al, 2006]: 1. Syntactic dimension, regarding the way the ontology is written2. Semantic dimension, concerning the consistent modelling of the ontology3. Functional dimension, related to the intended use of the ontology

SPARQL queries

SELECT ?object WHERE { rpms:Manager rpms:has_Category ?object }

SELECT ?object WHERE { rpms:agency_evaluate rdfs:range ?object }

Posgrado

EDON

Ontology Alignment Process

Posgrado

EDON

Ontology Aligment Activity in a study case

• A new ontology is created by establishing a set of correspondences between entities belonging to two different ontologies [Pavel et al, 2011]

• A correspondence Co is a 4-uple { id, e1, e2, r} where:• Id is the identifier of a given correspondence• E1 and e2 are entities of the first and the second ontology, respectively• r is the relation holding between e1 and e2

• So, the correspondence Co:= {id, e1, e2, r} asserts that the relation r holds between the ontologies entities e1 and e2

Posgrado

EDON

Ontology Aligment Activity in a new iteration

Some new terms

One-Man Research Project Co-Led Research Project Promoted Research Project

Project Co-Manager Promoted Project Team PhD Fellow

e1 (Entity of Ontology 1) r (Relation) e2 (Entity of Ontology 2)

Research Project ⊇ One-Man Research Project

Research Project ⊇ Co-Led Research Project

Research Project ⊇ Promoted Research Project

Researcher ⊇ PhD Fellow

Manager ⊥ Co-Manager

Manager ⊥ PhD Fellow

Posgrado

EXPERIENCES

From a study case

1. Importance of requirements grouping activityRequirements grouping activity has shown a deep influence in the efficiency of the overall ontology development process

2. More powerful conceptualization formalismAlthough using LEL has proven to be useful to facilitate the communication between the DEs and KEs, a more powerful formalism will improve the way complex business rules are expressed

Software development team highlighted the ability to adapt the overall software system to changing business requirements with minor modifications on the software procedural code

DEs showed interested on the possibility of embed its knowledge in the software system, participating in the process and evolving from knowledge to ontology in a smooth way

Posgrado

DISCUSSIONEDON partly draws from the knowledge developed in several fields

1. Using of competency questions from TOVE Project [Gruninger et al, 1995]

2. Definition of iterative life cycle

3. Definition of orthogonal activities4. Definition of collaborative work of DEs and KEs from Methontology

[Fernández-López et al, 1997]

5. Applying of the main principles of XP agile software methodology[Beck, 2000]

6. Using of Application Languages from LELs [Leite et al, 1993]

7. Using of a systematic process to evolve from LELs towards an ontology[Breitman et al, 2004]

Posgrado

DISCUSSION

Distinctive features

1. Development of an ontology intended as a structural conceptual model of an information system, encoding business rules in a declarative way

2. Aptness to intertwine the software and ontology development process

3. Ontology development through a phase model that takes advantage of the benefits of agile approaches

Posgrado

FUTURE RESEARCH DIRECTIONS

Will be focused on the following areas

1. Tracing a set of guidelinesfor the correct grouping of requirements will help the experts to improve the requirements selection process

2. Researching for a more powerful formalismfor ontology conceptualization with the aim to enable the expression of complex business rules in a high level of abstraction

3. Formulation of a systematic processto generate the ontology structure from such high level expressions

4. Acquiring of additional validation cases

Posgrado

Questions?Emiliano Reynares, María Laura Caliusco, María Rosa Galli

[email protected]@frsf.utn.edu.ar

[email protected]

CIDISI UTN FRSF