Deborah, Matthew, Ahsan, Steve, Ali, Nicola, and Jack, (01)
DMcP:
> In your view is it even possible for model-driven exchange environments
> to succeed without including ontologies? (02)
MW:
> I would argue that databases are willy nilly ontologies, since they
> make statements about the sorts of things there are, and some rules
> that govern them. Perhaps more importantly a database can be a very
> suitable implementation environment for an ontology, depending on
> what your purposes are. (03)
I agree with Matthew. In the 1970s, DB designers were discussing
very similar issues about ontology that we are talking about here. (04)
Type hierarchies, E-R diagrams, and Petri nets were used in the 1960s,
they were adopted by the DB community, and they are part of the UML
collection. In fact, UML diagrams are probably the most widely used
notation for ontologies on planet earth. UML plus OCL (the object-
constraint language) provides a *superset* of OWL, but in a much more
readable notation. (05)
The programming community is already familiar with UML diagrams,
which provide representations for type hierarchies, for the type
constraints and cardinality constraints on relations, for time
dependencies in activity diagrams, etc. If more expressive power is
needed, UML also includes OCL as a general-purpose notation for FOL. (06)
That is far more expressive power in a far more readable format
than OWL. You can translate any OWL ontology to UML, but not
vice-versa. (07)
AM:
> What do you think about SKOS-XL instead OWL for building ontology? (08)
I have no objection to anybody using whatever tools they find useful.
But it shows that OWL is a very difficult language to learn and use
effectively. Much simpler languages supplemented with diagrams
would be very attractive to many users. (09)
SW:
> Would that I had a nickel for every time I've seen someone misinterpret
> a "controlled English" sentence. (010)
I'd be delighted to take that bet, provided that you give me a penny for
every time I've seen somebody misinterpret a statement in some formal
language (logic, programming language, etc.). (011)
Please note that COBOL is a rather poor example of what can be done
with English-like syntax, but it was the most widely used programming
language during the second half of the 20th century. (012)
SW:
> My conclusion, then, is that end users are likely to understand
> the benefits of ontologies well before programmers. (013)
Programmers and database administrators understood the need for
ontologies since the 1970s. But they called them conceptual schemas,
structured analysis and design, etc. (014)
SW:
> The model in question is IDEF1-X. Information exchange is based
> on database replication... (015)
That's ontology! Note Matthew's comment and my response. (016)
AH:
> The argument I've used (with limited, but notable success) with the
> programmers around me, is that an ontology can also serve as a
> contract between the software development team and each module. (017)
I agree. And the people who were designing software development
tools in the 1970s used very similar arguments. The only missing
jargon was the word 'ontology'. Instead, they used terms like
'specification' or 'conceptual schema' or even 'IDEF1X'. (018)
AH:
> Fleshing out these different roles would be instrumental in helping
> focus and identifying the different types of supporting arguments. (019)
MB:
> Another way of framing this is that every application has an ontology
> anyway. The question is how it is framed, if at all. Are the meanings
> of terms resident only in the head of the developer, or in some logical
> model with written term definitions (weak semantics) or in a formal
> language which > grounds the meanings of terms with reference to some
> logical formalism? (020)
I agree with both of those statements. And I encourage anybody who
has been using OWL to take another look at UML. For most of what
they do with OWL, they could specify much more clearly with UML. (021)
SW:
> It's probably better to think of every application having multiple views
> of data. The view that is presented to the user may differ significantly
> from the view that's in memory, which in turn may differ from the view
> that's persisted. (022)
I strongly agree. And that's another argument for UML as a better
ontology language than OWL. The various diagrams give you multiple
views of the subject. But OWL is designed to enforce tunnel vision. (023)
SW:
> The case to make is that the OWL model of information is more "natural"
> than the relational model, so the application developer spends less time
> and effort translating a business model to OWL than to SQL. (024)
SQL happens to be a very primitive version of the relational model.
Ted Codd was not happy about the SQL version of relational semantics.
In 1979, Codd and Date made a strong case for adding a type hierarchy
in the RM/T extensions. In fact, I assumed a type hierarchy in my
first published article on conceptual graphs in 1976: (025)
http://www.jfsowa.com/pubs/cg1976.pdf
Conceptual Graphs for a Database Interface (026)
NG:
> Deciding how much effort to put in developing a particular ontology
> is a crucial choice, and it is very important to distinguish the cases
> where a proper ontological analysis pays off, and is indeed a crucial
> aspect of success, from those where a "lightweight" approach is sufficient. (027)
I agree. In fact, there is a great deal of informal analysis that must
be done before it's possible to write any kind of formal specification. (028)
JR:
> How about engaging them in a survey to estimate the cost of "IT Babel"
> in their respective enterprises? We might even mention the trillion
> dollar elephant in the room --- insecure systems. (029)
That is indeed a serious problem. People have been talking about it
since the 1970s. The only thing new is that the word 'ontology' has
been thrown into the pot. But talking has not solved the problems. (030)
John (031)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2011/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011
Community Portal: http://ontolog.cim3.net/wiki/ (032)
|