Against Idiosynchrasy in Ontology Development by Barry Smith
…and responses from Matthew West
Discussion on
Barry Smith- Julian Park Distinguished Professor of Philosophy in the Department of Philosophy at the
University at Buffalo
- Research Director of the Institute for Formal Ontology and Medical Information Science (IFOMIS).
- Smith’s current research focus is ontology and its applications in biomedicine and biomedical informatics, where he is working on a variety of projects relating to biomedical terminologies and electronic health records.
Source: http://org.buffalo.edu/rarp/smith_vita.html
Matthew West
If Tim Berners-Lee is the father of the internet, you could say Matthew West is the father of
ISO15926 (at least Part 2)
- Worked for Shell between 1978 and 2008
- Much of his work has been in the development of standards for information management (particularly engineering)
- Has been involved in PISTEP, POSC Caesar, EPISTLE, ISO TC184/SC4, BSI AMT/4, IEEE-SUO, KnoW and ONTOLOG
Source: http://www.matthew-west.org.uk/
The “possible individuals”
Background
ISO 15926 as an Upper Ontology?- Matthew West co-authored a paper in 2007 proposing ISO 15926 be put forward as an upper ontology.
- From this perspective, Barry Smith argues against this idea by pointing out a number of ‘defects’ in the design of the data model, including:
- Terminological confusions
- Failure to adhere to sound ontology construction principles
What is all the fuss about?
A namespace is a class_of_arrangement_of_individual where the class_of_whole and class_of_part are members of class_of_information_representation and the part is the most significant part of the whole that is the namespace.
Source: ISO 15926-2
Example: ISO 15926-2 definition for namespace
• The terms comprising the taxonomy are domain independant
• Example terms:
• SingleValuedRelation (SUMO)
• Process (SUMO,DOLCE)
• Class_of_biological_matter (ISO 15926)
• ISO 15926 is sufficient to model multiple domains:
• Example domains:
• Information Management (Libraries, Museums)
• Engineering (Design, Construction)
• Operations and Maintenance
• Chemical, Medical, Physical systems…
So on the surface ISO 15926 could be considered as an Upper ontology
What makes an Ontology an Upper Ontology?
• According to Smith, ISO 15926 breaches a number of principles (both as an ontology and an upper ontology)
• The principal of openness
• As an ISO standard, people have to pay to get hold of the specifications comprising the standard
• The principle of intelligible definitions
• E.g. A class_of_cause_of_beginning_of_class_of_individual is a class_of_relationship that indicates that a member of a class_of_activity causes the beginning of a member of a class_of_individual.
• The principle of non-circularity
• E.g. An <integer_number> is an <arithmetic_number> that is an integer number.
Barry Smith’s Principals
• B.S. Response: On the issue of classes named class_of_X and class_of_class_of_X
• M.W. Response: The data model is a specification for the database to hold any extension to the ontology. Thus entity types are required for instances but not the X’s
• No useful tools available to map domain specific data to iso 15926
• M.W. concedes to a couple of minor issues
• Explanation missing in documentation
• Typo’s in documentation
• These seem to be the only ground that M.W. concedes, however for the rest M.W. stands by his standard
Response to Against Idiosyncrasy in Ontology Development
• Model elements created to suit implementation rather than purpose
• Complicated and confusing naming for entities and types
• E.g. class_of_cause_of_beginning_of_class_of_individual
• Deals with class and metaclass at the one modelling layer
• Definitions written using subjective wording
• Use EXPRESS diagrams that are mind-bogglingly confusing
• E.g. class of namespace!
• Design of standard is heavily influenced by implementation
• Basing design decisions to suit implementation language i.e. OWL
Why ISO 15926 is NOT an Upper Ontology
According to Smith, ISO 15926 is more like a coding scheme such as the Standard Algebraic Notation for Chess…which is not an ontology of chess.