Hi Cesar, Dan and others,

A good explanation Dan of the WWW reason for doing things 'semantically',
one aspect that you have not explained perhaps is that you can think of the
web as a giant database, full of facts. How to make those facts more
useful? A way is for statements to be given more precise meanings, so that
AI 'reasoners' can infer implicit connections, to turn the web into more of
a graph database (to give the power to the people, not just to Google et.
al.).

As you say, ontologies are the means of giving things more precise meaning
(i.e. semantics). An ontology is a conceptual model (like an ER diagram).
The language used for this is W3C Web Ontology Language (OWL)
<http://www.w3.org/2001/sw/wiki/OWL>.

The basis of the Semantic web is that ALL data can be expressed in a form
thing->relationship->value, a 'triple', officially expressed as
subject->predicate->object. e.g. I can make two statements
Fluffy->is-a->Cat and Fluffy->has-colour->Black. If a reasoner can know
that Fluffy is the same Fluffy in both cases, it will be able to infer that
Fluffy is a Cat who is Black. To be able to know that Fluffy is the same
'thing' in both cases URIs are used to identify 'resources' explicity. So
all things can be identified using a URI, usually a URL.  So too can
concepts, all of is-a, has-color and Black can be explicitly identified
using this same URI method. Resource becomes a very general concept.

So apart from web pages annotated via RDFa <https://rdfa.info/> there are
databases called triple-stores that store data in this decomposed manner.
The language for querying such databases is SPARQL. I think this is a
summary of what the course will cover.  Any database of any type can be
presented to the world as if it is a triple-store. As far as I know the
only place this has caught on is in the medical world, but I am no doubt
not well informed.

What relevance of this too Apache Isis? Well you nearly got it straight
off, that the Isis meta-model is a model that maybe can be exported as OWL.
I don't think so as its more of logical model than a conceptual one. I am
thinking the other way around, that a conceptual model in OWL, or something
else, could be the basis of the Isis meta-model.

My interest in all this is largely a hunch at this point. I am intrigued by
the idea of people being able to manage data via the web in a way that is
empowering. This has been possible via a desktop application called
Microsoft Access for several decades, why is it still so hard to do it via
the web? All we have still (correct me if I am wrong) is online
spreadsheets, spreadsheets are good-ish for analysis, they are not a
database and infact are essentially a physical model. Access gives people a
relational modelling tool, I think you should could give them a conceptual
model and let the system do the rest, as much as is possible.

The thing that led me to Isis was that half of this picture is already
done, the model to UI part, I thought maybe I could add the other half
(person to model)? Also the meta-model in Isis is a OO model, which is
nearly what you want perhaps? That is what I want to get a handle on via
this course.

Regards
Steve





On Tue, Nov 3, 2015 at 6:25 AM, Dan Haywood <[email protected]>
wrote:

> Thanks for the precis, Cesar... very interesting.
>
> Clearly in Isis we have a lot of the information in our metamodel that
> could be the input for a semantic web ontology; I'd appreciate
> understanding how this information is "exported" such that an Isis app can
> also be considered as a "semantic web app" (if that's even a term...)
>
> cheers
> Dan
>
>
> On 2 November 2015 at 15:59, Cesar Lugo <[email protected]> wrote:
>
> > Dan, I think this is an important and very interesting topic. I have some
> > basic knowledge of the semantic web concepts, and I have done some
> > significant prototyping using semantic web tools, just far from being an
> > expert or knowledgeable person. As far as I k now, the semantic web
> concept
> > was developed by the father of the www, and it adds semantics to the web
> > content, and to the web applications as well. This has also been called
> the
> > Web 3.0, and the semantic web technologies go from adding semantics to
> the
> > web pages, to build entire semantic applications.
> >
> > On the UI web pages side, it adds semantics embedded within the HTML with
> > semantic extensions, so a browser would be able to know more information
> > about the semantics of that page. For example, when you look at an
> amazon's
> > web page and find a book, you know that you are looking at a book, you
> know
> > what a book is, you know it's about growing apples, and you know some
> > numbers besides it are the different prices of the book, and the text
> > besides it is an overview of the book's contents. The browser, currently
> in
> > web 2.0, does not know any of those semantics, but just the page content
> in
> > terms of things like fields, controls and links. The semantic web lets
> the
> > browser infer all those semantics from the web page, buy using ontologies
> > and extended dictionaries (they are currently creating ontologies based
> on
> > Wikipedia content, but structured so the systems can understand and infer
> > meaning, not just content). The idea is that having the browser know the
> > semantics, it will help interact among other systems in a more semantical
> > way, like, for example, if you are interested in growing apples, then you
> > might also be interested in growing some kinds of organic food.
> >
> > Then, at the domain logic layer, you use ontologies (expressed in RDFs)
> to
> > represent your domain, which allows you to re-use existing ontologies for
> > common topics like a customer, a vendor, a company, the relationship
> among
> > people (like the FOAF friend of a friend ontology), a product, and more.
> > Then, there is a language called SPARQL, similar to SQL, but works at the
> > ontology level. All that and more lets you create your apps with semantic
> > content all over the app, not only at the UI. Your application will be
> more
> > knowledgeable about the semantics of the topics (domain objects) within
> it,
> > so you can use knowledge management methods to define and manage your
> > domain logic and your exposed content, and interact with other systems
> at a
> > semantic level.
> >
> > Cesar.
> >
> > -----Original Message-----
> > From: Dan Haywood [mailto:[email protected]]
> > Sent: Monday, November 2, 2015 2:52 AM
> > To: users
> > Subject: Re: Course of Interest
> >
> > Shall be interested to hear your thoughts on this... not a subject I have
> > any knowledge of.
> >
> > cheers
> > Dan
> >
> > On 2 November 2015 at 02:40, Stephen Cameron <[email protected]
> >
> > wrote:
> >
> > > Hello Apache Isis Community,
> > >
> > > I am about to start an free online course on Semantic Web
> > > Technologies. At the moment the connection to Isis is not clear, but I
> > > have ideas. I am doing this course to see if my ideas have any
> > > relationship to reality, specifically in regard to conceptual models.
> > >
> > > https://open.hpi.de/courses/semanticweb2015?locale=en
> > >
> > > Regards
> > > Steve Cameron
> > >
> >
> >
> > ---
> > This email has been checked for viruses by Avast antivirus software.
> > https://www.avast.com/antivirus
> >
> >
>

Reply via email to