[Top] [All Lists]

Re: [ontolog-forum] doing standards [was - Re: Webby objects]

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: John F Sowa <sowa@xxxxxxxxxxx>
Date: Thu, 22 Nov 2012 17:55:17 -0500
Message-id: <50AEAD55.5010506@xxxxxxxxxxx>
Peter, Andries, Godfrey, William, Brand, Ron, Rich, Amanda, Leo, David,    (01)

I can see that a lot of people are doing their email while waiting
for the turkey.  Happy Thanksgiving.    (02)

> I hope to see [the W3C] among the collaborators too, and solicit
> help from anyone here who can help get W3C leadership's attention on this
> on this initiative and put me in touch.    (03)

That would be great.  We should try to get everybody involved.    (04)

> Mainstream practitioners do not use ontologies, because they are oriented
> towards databases and not to messages and for that purpose they are educated
> to create 'data models' to design those databases.    (05)

I keep meeting mainstream practitioners with widely varying backgrounds.
Some have PhD's in philosophy and would love to do ontology.  But they
write application programs because they have to support their families.    (06)

> Data models are in fact only collections of 'fill-in the blanks'
> templates  with predefined classifications of their content.    (07)

I went down to my basement "stacks", where I have books and filing
cabinets of stuff I haven't looked at in years.  I dusted off books
and reports on data modeling from the 1970s.  The most depressing
point is that many of them were discussing the same kinds of issues
and proposals we're debating right now.  The main difference is that
they were using the term 'conceptual schema' instead of 'ontology'.    (08)

But I agree that most users want to "fill in the blanks" -- often
in their spreadsheets.  And we should design our ontology tools to
use those spreadsheets as input.    (09)

> ... the majority of OWL ontologies in use are basically just class/subclass
> and/or whole/part hierarchies with possibly one or two other constraints.
> We’ve built a couple of dozen like that.
> In the content/media industries it is now commonplace for standards and
> organizations to rely on complex sets of controlled vocabularies,
> usually involving hierarchies.    (010)

Yes.  That's why I said that the vocabularies in Schema.org are very
easy for developers to adopt.  They can be combined with UML diagrams
to specify type hierarchies, part-whole hierarchies, signatures of
relations and the type and cardinality constraints.    (011)

> The key point to me is that the fancy, experimental tools get in
> the way of business people who are the ones who have to own this stuff,
> and present risks for enterprise IT.  The most complete and effective
> low level domain ontologies have seen is usually done in EXCEL.    (012)

> +1    (013)

I'll add ++.  You can give a domain expert a hierarchy of vocabulary
terms from Schema.org to use as the labels on their Excel columns,
and they can use them immediately.  The IT department can use the
same labels, supplemented with UML diagrams (which they already know),
and upload the spreadsheets to their databases.    (014)

> Yesterday I was with a modeller from a large media organization who
> has developed an extensible seven-dimensional matrix of rights management
> categories containing hundreds of terms and permutations, guaranteed to
> grow and develop substantially on a regular basis and needing to map
> to complex vocabularies and schemas in other internal and external systems.    (015)

Yes.  Programmers and their managers understand the need for naming
conventions, and Schema.org with the Good Relations ontology gives
them something they can begin using immediately.  It is also in the
common intersection of any and every kind of logic and database system.    (016)

> It seems that far too much time is spent arguing about 'what the meaning
> of "is" is' and getting caught up in long discussions that often have
> more to do about different people ability to parse the English language
> than actually looking at what role Ontology can play in solving real
> world problems.    (017)

Those subtle issues are necessary for advanced tools.  But the hardest
part of formalizing large volumes of data is getting a good vocabulary
of terms and putting the data into any kind of structured format.  If
you get the data in a spreadsheet or a DB with a standard vocabulary
of labels, it can be mapped to any format of any kind.    (018)

> Consider two business partners that exchange data.  For example,
> a lender and a title company.  All they care about exchanging is
> information about the property getting the loan.  All additional
> ontology structuring is irrelevant.  They certainly wouldn’t want
> to use the SAME ontology because they are in different businesses
> and have different concerns.  The only ontology they would care
> about relates to properties on which they lend, and not on concerns
> about the definition of a field or river or mountain.    (019)

I agree.  Very underspecified labels at the level of Schema.org are
what they need for this task.  Detailed axioms and definitions would
be worse than useless.  They would almost certainly have incompatible
or inconsistent definitions that would not help in any way.    (020)

>> What is missing from our tools, techniques, and logics?    (021)

> Practical value, ease of use, simplification like that used by Google,
> yet not pressed into the consumers’ faces.    (022)

If the tools developed by and for the Semantic Web had that level of
simplicity, they would have become part of the mainstream years ago.    (023)

>> What methods work in mainstream development?    (024)

> Software engineering, including HTML, SQL, XML, and focused development
> to meet specific requirements, increment by increment.  What extra value
> does ontology bring, if any? -- very little demonstrated value beyond
> Dublin Core has been seen in the industry.    (025)

Yes.  The mainstream tools should be augmented with tools that use
Schema.org + Good Relations + Dublin Core + vocabularies from
every branch of science, engineering, business, medicine, law, etc.    (026)

> Theory has no practical value, though it has immense scientific value.    (027)

More precisely, theories from science are essential for engineering,
and engineering has immense practical value.  But don't expect ordinary
users to learn the theory.    (028)

> Controlling language -- setting and using naming standards, for example,
> -- is *much* harder than building and using shared ontologies.    (029)

Yes.  That is the point that Martin Hepp emphasized about his work on
Good Relations.  Doug Skuce also observed that it was the most essential
step in any IT project.  He said that tools for writing and managing the
vocabulary are essential.  An example is his Fact Guru, which can be
used by domain experts who have no special training in logic, ontology,
or programming.    (030)

> We need a good understanding of what ontology features matter to what
> kinds of projects: that is, an understanding of what it means for an
> ontology to be suitable for an application, fit for a particular use case.    (031)

Much more work on human factors is needed, but I don't believe that any
KR language designed by and for AI experts can be used by programmers,
web masters, and other IT professionals.    (032)

> Until we build that kind of understanding -- at least a start -- some
> people who could benefit from ontology applications will continue
> to balk at the lack of clarity, while other people will go ahead but
> invest in ontology that is mismatched to the requirements.    (033)

I don't believe that we (the AI community) can train domain experts,
programmers, IT professionals, and data-entry clerks all the niceties
of logic, ontology, and knowledge representation.  The ones who have
the most training in their profession are probably very highly paid.
Except for unemployed philosophers, their time is too valuable to
waste on learning KR methodologies.    (034)

But they can use spreadsheets, controlled vocabulary hierarchies (at the
level of Schema.org), UML diagrams, and controlled English.  There are
also many kinds of automated and semi-automated tools that can check
for various subtle issues that could cause problems in reasoning.    (035)

> A good ontologist can, in fact, come in and build something and make
> a difference in two weeks.BUT this assumes that it is clear what
> the ontology is for, how it will be used, what requirements come from
> downstream, and what sources are to be regarded as authoritative,
> AND that the ontologists is provided access to those sources.    (036)

> Sure, Amanda, and that’s why I (and we) advocate using natural language
> vocabularies that are linked/mapped to ontologies. This was a hard lesson
> learned  (initially, by others, before my time) in the DoD in the
> early 1990s, and that I personally experienced in the metadata wars
> of the 2000s, where people will fight to the death to include their “words”,
> mistaking these for the concepts behind them.    (037)

I agree.  But a lot can be done with automated and semi-automated tools.    (038)

> I would argue that *** IF *** ontology work products were as accessible
> as a spell checker then my personal work products would be more accurate
> & professional.
> Plus, I get to train my ontology catcher one word at a time.
> I would obviously advocate for a smarter spell checker that groks context    (039)

I very strongly agree.  And tools like that are coming.  It's much more
realistic to design tools that can extract ontology from ordinary
language than to expect ordinary users to learn ontology.    (040)

At VivoMind, we take documents (doc files, PDF files, email, etc.) and
extract the named entities and tentative ontologies.  It's even easier
to derive ontologies from formal programming languages and database
specifications.  Then we use that ontology to analyze the manuals,
reports, and emails about that software to determine the words and
phrases that refer to the software components and the data items.    (041)

This has produced some very good results, which can be used by IT
personnel who have no training in ontology.  See slides 111 to 155
of http://www.jfsowa.com/talks/goal.pdf    (042)

John    (043)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (044)

<Prev in Thread] Current Thread [Next in Thread>