Introduction

This REPRO handbook is intended for newcomers and experts to ontology engineering (OE) in order to learn and share about its practices and backgrounds.

"The stuff of the world is mind-stuff "

-- Arthur Eddington

Sir Arthur Stanley Eddington (28 December 1882 – 22 November 1944) was an English astronomer, physicist, and mathematician of the early 20th century who did his greatest work in astrophysics. He was also a philosopher of science and a populariser of science. This quote originates from his book The Nature of the Physical World.

With this quote, Eddington stressed that, in order to measure the quantities, one has to interact with the world with his or her consciousness. He argues that physical science presupposes a background outside its own scope, and that this background, ‘the substratum of everything’, is mental in character. Eddington calls this background ‘mind-stuff’. Our interpretation is that ontology represents that very mind-stuff, or, in more contemporary terms, a conceptualisation.

Purpose of the REPRO Handbook

TNO's Data Science department produces semantic standards in order to facilitate semantic interoperability between different stakeholders, intra- and inter-organizational alike. Where we historically standardised on the level of schemata and message structures, we are now bringing that semantic standardisation work onto a next level of abstraction. The objective of such higher level of abstraction is, first and foremost, to consider semantics a separate concern that exists independently from any other artefact, technology or representation. Separation of concerns is a fundamental principle that allows qualities to emerge, such as reusability and evolution, accuracy or usability, and many other "-ilities" that are similar to software qualities. We, thus, now standardise on the level of ontologies, producing semantic interoperability that is fully independent from the communication paradigm that takes care of establishing technological interoperability. Consequently, an appropriate ontology engineering standard is required that: fits our way of working; is founded on proven ontology engineering methods; provides and discusses best practices on very practical issues that are encountered; and identifies patterns for specific cases. This Handbook, although still work in progress, is meant to provide exactly that.

More and more individuals from our department are involved in ontology engineering. In that process we acknowledge that engineering is an art that involves many design decisions about how to best achieve a certain goal. Regarding semantics, that goal is a particular way to carve out a specific part of reality such that the actors and stakeholders can effectively engage in a conversation that follows some of Grice's conversational maxim's [Grice75]:

(i) make your contribution as informative as is required (for the current purpose of the exchange);
(ii) do not make your contribution more informative than is required.

Indeed we standardise semantics for a particular purpose, but we acknowledge the following quote from Janowicz:

“The successful standardisation of protocols made us believe that we should also standardise meaning on the Web. This is a fundamental misconception.” [Janowicz12].

This quote originates from the observation that semantic conventions consider semantic heterogeneity a bug instead of a feature necessary to achieve semantic accuracy. Semantics is a representation of some part of the world, viewed from a particular perspective of use. This is fundamental to the domain users when building semantics and getting an understanding. And equally important, that particular perspective is just one out of many equally legitimate ones. In other words, there is no single truth when it comes to semantics. Consequently, there is no semantic standard that applies universally which indicates that for every application a new semantic standard needs to be developed. Fortunately, organisations and networked businesses have the tendency to be rather conservative on going about their business: the why and what of their operations remain pretty much consistent over the years. This results in a rather stable view on their application domain, and hence their particular perspective on their particular domain of application do not change that much. Although their semantics will evolve over time, their changes are not as impactful and frequent as occur in ICT. In such environment a (mutual) semantic standard is quite a valuable asset the reuse of which will reduce costs significantly.

In conclusion, we need to master how to engineer ontologies in order to make the right decisions at the right moment in the process in order to represent the right things from the right perspective with the goal to engage in the right conversation. And it is our aim that the REPRO handbook provides ourselves with the right material to achieve just that.

Contents of the REPRO Handbook

The handbook identifies and describes the three main aspects that we deem necessary to address in order to achieve a standard engineering process. These three aspects emerge as the three Parts of this handbook:

Part I - a principled approach to engineering an ontology, as elaborated or thin as we deem necessary and sufficient.
Part II - a discussion of best practices, very pragmatic, reflecting the problems that we encounter and have discussed during the meetings;
Part III - during our discussions we have seen the need for patterns, singled out from engineering structures that resolve the same type of problems and that are considered to improve the quality of the total ontology.

All aspects are sourced from our regular discussions, our experiences from projects and the theories that can be found in literature. At the same time, in our opinion a method or approach should not result in a dogma or otherwise heavy procedure that exist for having a procedure alone. First and foremost, the existence of the Handbook is grounded in facilitating the engineer to produce an ontological artefact on a quality level that is deemed necessary by the context of its use and the scope of its lifecycle. The Handbook, therefore, does not prescribe a fixed procedure with obligatory steps that cannot be adapted. Instead, it must be complete enough to describe whatever is required throughout ontology engineering; in its descriptions, it must provide sufficient detail to guide the engineer in making the right decisions, but not more than that; and it must show how the results of that engineering process should be validated. Finally, it should show why certain elementary tasks belong to one phase and not another, combining tasks into logical phases with clear handovers between them. Based on such footing, every engineer can decide upfront what is required for the project at hand, and design the most appropriate process for it.

Overview of the REPRO Handbook

The REPRO handbook contains three parts: the REPRO Method, the REPRO Best Practices, and the REPRO Patterns. Besides that, a Closure consists of the literature references and relevant links.

Introduction How to become an ontology expert?

Part I - The REPRO Method

Chapter 1: Overview of the REPRO method Chapter 2: Preparatory Phase 2.1: Purpose & Intended use 2.2: Motivating business scenarios 2.3: Glossary of terms 2.4: Application requirements 2.5: Test definitions Chapter 3: Conceptual Phase 3.1: Competency Questions 3.2: Test Case Definitions 3.3: Ontology modularisation 3.4: Conceptual Modelling 3.5: Axioms Definition 3.7: Ontology Formalisation 3.8: Query & Test Data Formalisation 3.9: Ontology Verification Chapter 4: Computational Phase Chapter 5: Ontology Publication

Part II - The REPRO Best Practices

Chapter 1: Identifiers Chapter 2: Synonyms 3.1: Deciding on reuse of existing ontologies Chapter 4: Codelists Chapter 5: Generic vs specific

Part III - The REPRO Patterns

Still empty.

Closure

All references link to the central Literature section

Literature Relevant links

Last updated 3 years ago