Use of ontologies in software development

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

Use of ontologies in software development

NEHEMIAH MAVETERA

Dear All

Is there anyone who has used ontologies in software development. I mean at run time.  Please furnish me with any projects that you have done in this regard.

My problems are that, as a systems developer, analyst you name it, what should I look for in a software development project that can be enhanced using ontologies.

A typical example is " Use of ontologies in software development for agent mediated electronic marktes" . How do I make use of ontologies at run time?

Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA; MSAICSIT; (SA))
Senior Lecturer- Information Systems
North West University
www.uniwest.ac.za
Tel: +27183892143
Fax: +27183892090

Can we isolate reason, and is it, so regarded, an independent source of concepts and judgements which spring from it alone? From  I. Kant, Critique of Pure Reason, A305-B362
>>> [hidden email] 07/20/06 2:13 AM >>>

Well, probably someone else could answer that better than me, but I feel
there are some comments I would venture to make.

An ontology for a neuron should imply an ontology for a cell. So, if
properly designed, should contain a potential for a complete cellular and
subcellular ontology. This is, to the best of my knowledge, what has been
done in the Formal Model of Anatomy. In the context of our project, which I
mentioned earlier, we have used Digital Anatomist for modelling anatomy as
it pertains to surgical procedures. So, although I am streching the concept,
in a certain sense an ontology for cells is part of our project. We have not
as yet started with the ontology of laboratory procedures, but when we reach
that point I am sure we will be using the cell ontology which you indirectly
mention, proably drawn form Digital Anatomist.

Hope my comments have helped.

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf
> Of sridevi polavaram
> Sent: Wednesday, July 19, 2006 2:29 PM
> To: [hidden email]
> Subject: [protege-discussion] Re: I want some solid real
> world examples!
>
> Alan,
>  Let's say that i have an owl description of the ontology for
> Neuron (cell structures in brain that processes information).
> Now what? how does this owl file help me? according to my
> understanding, i know that having an owl file is like
> specifying the semantics of Neuron concept.
> Tell me what's the first use that comes to your mind seeing
> an owl file. you can probably use ur own example.
> -Thanks
> Sri
>
> On 7/18/06, Alan March <[hidden email]> wrote:
> >
> >
> > Yep. In our case its SQL Server. What I meant by "RDF" is
> that we use
> > a table which contains the info that would otherwise be
> contained in
> > the RDF/OWL File. Sorry, I loosely called it "RDF table" but that's
> > just a local designation we use to refer to that table (the "rdf"
> > table"...granted: it does no service to proper understanding of the
> > concept...true reification in the literal sense of the
> term). You can
> > actually call it any name in the JDBC implementation of Protégé.
> >
> > Regarding the use of "list": An adjacency list is a manner
> in which to
> > represent a graph structure and it can be implemented in a
> database table.
> > It's a classical implmentation of graph structures. Joe Celko has
> > written extensively on that, using ANSI-SQL.
> >
> > Finally, there is a lot of literature on how to map (that is,
> > represent) RDF structures to a RDBMS. The authors of the
> JDBC backend
> > made it simple and I actually think they did it very well. Its
> > supposed inefficiency can be dealt
> >
> > A word of advice: if you use SQL server as a backend, don't use
> > Microsoft JDBC driver. I think there's an issue regarding the Java
> > runtime, but it simply hangs when you try to use a reasoner, with a
> > message regarding "column 6" which I frankly did not
> research into. I
> > just dumped the driver.
> >
> > Regards,
> >
> > Alan
> >
> > > -----Original Message-----
> > > From: [hidden email]
> > > [mailto:[hidden email]] On Behalf Of
> > > sridevi polavaram
> > > Sent: Tuesday, July 18, 2006 3:23 PM
> > > To: [hidden email]
> > > Subject: [protege-discussion] Re: I want some solid real world
> > > examples!
> > >
> > > Alan,
> > > when you say database backend, i am assuming it is some
> RDBMS like
> > > mysql, oracle etc., but you are referring it as RDF table
> and a list
> > > structure! so what kind of DB is this! can u maintain owl or RDF
> > > DBs!
> > > what is the size limit of these DBs? and how exactly are you
> > > achieving data storage and retrieval methods!
> > > -regards,
> > > Sri.
> > >
> > > On 7/18/06, Alan March <[hidden email]> wrote:
> > > >
> > > >
> > > > Well, one of the first problems we run into is OWL itself.
> > > Although it
> > > > is very powefull in expressivity, we believe it is not
> > > efficient as a
> > > > storage and retrieval repository (indeed, it was probably
> > > not designed
> > > > with efficiency in mind, and in that sense it reminds
> me of my old
> > > > Prolog days).
> > > > So we used the JDBC database backend. The database backend
> > > is not that
> > > > efficient in itself (and, again, its creators mention in the
> > > > documentation that efficiency was not one of their
> > > requirements) but
> > > > it is extremely elastic. Stated simply, the DB backend is a
> > > (sort of)
> > > > adjacency list with the particularly interesting
> > > characteristic that
> > > > the list is double linked.
> > > > From what I gather, one important difference with the pure RDF
> > > > approach is that nodes are loaded on demand, and there
> is a slight
> > > > performance issue there, but then, it allows for managing large
> > > > ontologies. So all design is managed using Protégé as
> the front-end.
> > > >
> > > > The advantage of the database backend is that it makes all the
> > > > elements of the ontology easily accesible for other
> > > applications when
> > > > these applications aren't involved with the design task itself.
> > > > Example: the multiple classifications built by reasoners such as
> > > > Fact++ may be accessed from web pages and applications using
> > > > Fact++ simple
> > > > approaches such as treeviews and the like, according to user's
> > > > different classification requirements (a CFO is interested
> > > if a cost
> > > > based classification of procedures, whilst a chief medical
> > > officer is
> > > > interested in a body systems classification of procedures).
> > > > Another
> > > > example: extracting dimensions for a warehouse. And of course
> > > > there are more.
> > > >
> > > > Most of these tasks are carried out by writing queries to the
> > > > database. We use stored procedures, which are included
> in the same
> > > > database  that contains the main "RDF" table. The pattern
> > > > generally used is a self linked table for accessing the
> adjacency
> > > > list schema
> > > >
> > > > Don't know how much "state of the art" this is, but it
> works for us.
> > > >
> > > > Regards,
> > > >
> > > > AM
> > > >
> > > > > -----Original Message-----
> > > > > From: [hidden email]
> > > > > [mailto:[hidden email]]
> On Behalf Of
> > > > > sridevi polavaram
> > > > > Sent: Tuesday, July 18, 2006 9:35 AM
> > > > > To: [hidden email]
> > > > > Subject: [protege-discussion] Re: I want some solid
> real world
> > > > > examples!
> > > > >
> > > > > Alan,
> > > > >  Thankyou very much for explaining things so
> patiently. So, how
> > > > > exactly are you using the ontology that you have tailored
> > > for your
> > > > > purpose at the application level? are you mapping the
> ontology
> > > > > properties as attributes at the database-level?
> > > > > if so, what are the other kinds of mapping that u do?
> > > > > are u using any logic based programming languages? How is
> > > the rich
> > > > > annotation that's inherent in ontologies is helping
> you at the
> > > > > application level, apart from the fact that it serves as a
> > > > > reinforcement knowldge base for humans. how are you
> > > extracting and
> > > > > storing information using ontologies?  can you say that
> > > ontology is
> > > > > crucial for your application and that it's not
> replacable by any
> > > > > other design methodologies.
> > > > > I am sorry but my mind is full of questions...just can't stop
> > > > > the flow..i am reading all the related literature that's
> > > > > available (white papers, tools, theroies, real examples,
> > > > > technology
> > > etc.,) but
> > > > > none of them help me converge my understanding.
> > > > > -regards,
> > > > > Sri
> > > > >
> > > > > On 7/17/06, Alan March <[hidden email]> wrote:
> > > > > >
> > > > > >
> > > > > > > > Wow SNOMED and UMLS are supposingly success stories for
> > > > > > > > ontology
> > > > > > > applications. and they failed to mee your requirements!!
> > > > > > > then the whole point of "sharing knowldge" seems
> skeptical.
> > > > > > > So are u designing ontologies from scratch for your HMO
> > > > > application!
> > > > > > > everyone is apparently having there own ontology
> > > > > >
> > > > > > The trouble we faced with SNOMED and UMLS is they
> > > appeared to us
> > > > > > as incomplete as far as attributes (relations) are
> > > concerned: too
> > > > > > few, failing to capture the richness of possible relations.
> > > > > > And then of course, the mistakes found in them, which Barry
> > > Smith and
> > > > > others have
> > > > > > mentioned in their papers. Also, much of the terms we had
> > > > > to use were
> > > > > > highly local and neither SNOMED nor UMLS offered a clear
> > > > > > method for extensions. Although SNOMED has
> something on those
> > > > > > lines,
> > > > > the problem
> > > > > > of its "ontological incompleteness"
> > > > > > and licensing questions represented a drawback for us. But
> > > > > then again:
> > > > > > both
> > > > > > SNOMED and UMLS have been extremely usefull to us
> as a sort of
> > > > > > knowledge repositories, but we just could not use them
> > > > > "out-of-the-box".
> > > > > >
> > > > > > Actually, we didn't start from scratch: we reutilized
> > > much of the
> > > > > > publically available Galen files, and modelled our ontology
> > > > > following
> > > > > > an upper level ontology. For anatomy we used Digital
> > > Anatomist as
> > > > > > published in UMLS. We tried to use as much as possible from
> > > > > > several ontologies we found to be suitable to our purposes.
> > > > > > But
> > > > > yes, we had to
> > > > > > reinvent the wheel in some circumstances. In any case, I am
> > > > > not quite
> > > > > > sure if we will ever manage to arrive at a universally
> > > > > > accepted ontology. Probably at the upper and middle levels
> > > > > > this will be achieved, but I doubt it will work for "bottom"
> > > > > > levels.
> > > > > >
> > > > > > Regarding success stories: well, I guess SNOMED is a
> > > > > success story in
> > > > > > a certain sense, but I am not quite sure if it can be
> > > > > called that way
> > > > > > as far as ontological principles are concerned. Not to deny
> > > > > its value:
> > > > > > it is rich in content. Worth mentioning is the fact
> > > that we didn't
> > > > > > want to consider terminologies which weren't in the public
> > > > > domain or commercial in nature.
> > > > > > I
> > > > > > personally feel that medical terminologies such be
> built in a
> > > > > > collaborative manner, probable on the same lines as Linux
> > > > > was made (by
> > > > > > having different groups managing different parts of the
> > > domain).
> > > > > > In that sense, I feel that Rector's advice regarding the way
> > > > > ontologies
> > > > > > should be built (by
> > > > > > "untangling") allows for this. I am relly sorry that the
> > > > > Galen project
> > > > > > didn't continue. Looking back on it, I think it is
> > > still the best
> > > > > > thing we have seen in all these decades of discussion on
> > > > > terminologies
> > > > > > and how they should be built and maintained.
> > > > > >
> > > > > > >
> > > > > > > In that sense, I find that ontologies are "the" techique
> > > > > for design,
> > > > > > > > development and maintaince of so-called controlled
> > > > > terminologies.
> > > > > > >
> > > > > > >
> > > > > > > > yes, i would say, that even i am convinced that at
> > > the level
> > > > > > > > of
> > > > > > > vocabularies. like a dictionary sort of thing. the
> > > wikipedia for
> > > > > > > encyclopedia also uses ontologies!!
> > > > > > > Thanks a lot for giving me some idea about the real workd
> > > > > scenario.
> > > > > > > Best,
> > > > > > > -Sri
> > > > > > >
> > > > > > > OF course,
> > > > > > > > in the process of using ontologies one quickly realizes
> > > > > > > > the
> > > > > > > fact that
> > > > > > > > they could probably be usefull for analysis and design
> > > > > of complex
> > > > > > > > systems, and I personally feel that they could
> > > > > ultimately replace
> > > > > > > > other tools such as UML and the like. This, of
> course (and
> > > > > > > at least as
> > > > > > > > far as my comprehension goes), still requires much work
> > > > > regarding
> > > > > > > > methodology and probably software tools. But I feel that
> > > > > > > there is more
> > > > > > > > to ontologies than what meets the eye of the casual
> > > > > user. Example:
> > > > > > > > I've been following developments in HL7 v3 and am every
> > > > > day more
> > > > > > > > convinced that v3 should be redesigned using readily
> > > > > > > available tools
> > > > > > > > such as Protégé (I don't think it shoud be
> thrown away as
> > > > > > > others have
> > > > > > > > suggestes, as the RIM contains a lot of useful domain
> > > > > > > information, but
> > > > > > > > the methodolgy user does not seem adequate).
> > > > > > > >
> > > > > > > > So, I personally feel that ontologies and Protégé-type
> > > > > tools could
> > > > > > > > turn out to be a small revolution in systems
> > > design. Should my
> > > > > > > > idea prove right, we are now in a sort of
> pre-paradigmatic
> > > > > > > situation as far
> > > > > > > > as application of ontologies is concerened.
> > > > > > > >
> > > > > > > > Best regards
> > > > > > > >
> > > > > > > > > -----Original Message-----
> > > > > > > > > From: [hidden email]
> > > > > > > > > [mailto:[hidden email]]
> > > > > On Behalf Of
> > > > > > > > > sridevi polavaram
> > > > > > > > > Sent: Monday, July 17, 2006 12:52 PM
> > > > > > > > > To: [hidden email]
> > > > > > > > > Subject: [protege-discussion] I want some solid real
> > > > > > > world examples!
> > > > > > > > >
> > > > > > > > > Hello All,
> > > > > > > > >  Please help me find an answer to my questions. I
> > > > > suppose some
> > > > > > > > > of you might also have the same questions. so let's
> > > > > brainstorm.
> > > > > > > > > I totally understand what ontology is
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > >
> > >
> --------------------------------------------------------------------
> > > > > > > --
> > > > > > > > --- To unsubscribe go to
> > > > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> ------------------------------------------------------------
> > > > > > > --
> > > > > > > -----------
> > > > > > > To unsubscribe go to
> > > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > >
> --------------------------------------------------------------------
> > > > > --
> > > > > > --- To unsubscribe go to
> > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > >
> > > > > >
> > > > >
> > > > > --------------------------------------------------------------
> > > > > -----------
> > > > > To unsubscribe go to
> > > > > http://protege.stanford.edu/community/subscribe.html
> > > > >
> > > > >
> > > >
> > > >
> > >
> --------------------------------------------------------------------
> > > --
> > > > --- To unsubscribe go to
> > > > http://protege.stanford.edu/community/subscribe.html
> > > >
> > > >
> > >
> > > --------------------------------------------------------------
> > > -----------
> > > To unsubscribe go to
> > > http://protege.stanford.edu/community/subscribe.html
> > >
> >
> >
> ----------------------------------------------------------------------
> > --- To unsubscribe go to
> > http://protege.stanford.edu/community/subscribe.html
> >
> >
>
> --------------------------------------------------------------
> -----------
> To unsubscribe go to
> http://protege.stanford.edu/community/subscribe.html
>

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html


-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

Reply | Threaded
Open this post in threaded view
|

Re: Use of ontologies in software development

NEHEMIAH MAVETERA

Dear All

Please is there any one who has projects where you have managed to map Existential Graphs(EG), Conceptual Graphs(CG) through predicate calculus and mereology  on to a software application

In fact as a systems analyst, I can capture systems and business requirements as CGs,and EGs and the onology of a system can thus be developed. Now, the problem is to convert it to machine readable form understood by the computer as a software application.

My research focuses on the need to move from explicit to implicit programming using the theoretical concept of ontologies.

Please assist.

Regards

nehemiah

Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA; MSAICSIT; (SA))
Senior Lecturer- Information Systems
North West University
www.uniwest.ac.za
Tel: +27183892143
Fax: +27183892090

Can we isolate reason, and is it, so regarded, an independent source of concepts and judgements which spring from it alone? From  I. Kant, Critique of Pure Reason, A305-B362
>>> [hidden email] 07/20/06 4:07 PM >>>

Dear All

Is there anyone who has used ontologies in software development. I mean at run time.  Please furnish me with any projects that you have done in this regard.

My problems are that, as a systems developer, analyst you name it, what should I look for in a software development project that can be enhanced using ontologies.

A typical example is " Use of ontologies in software development for agent mediated electronic marktes" . How do I make use of ontologies at run time?

Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA; MSAICSIT; (SA))
Senior Lecturer- Information Systems
North West University
www.uniwest.ac.za
Tel: +27183892143
Fax: +27183892090

Can we isolate reason, and is it, so regarded, an independent source of concepts and judgements which spring from it alone? From  I. Kant, Critique of Pure Reason, A305-B362
>>> [hidden email] 07/20/06 2:13 AM >>>

Well, probably someone else could answer that better than me, but I feel
there are some comments I would venture to make.

An ontology for a neuron should imply an ontology for a cell. So, if
properly designed, should contain a potential for a complete cellular and
subcellular ontology. This is, to the best of my knowledge, what has been
done in the Formal Model of Anatomy. In the context of our project, which I
mentioned earlier, we have used Digital Anatomist for modelling anatomy as
it pertains to surgical procedures. So, although I am streching the concept,
in a certain sense an ontology for cells is part of our project. We have not
as yet started with the ontology of laboratory procedures, but when we reach
that point I am sure we will be using the cell ontology which you indirectly
mention, proably drawn form Digital Anatomist.

Hope my comments have helped.

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf
> Of sridevi polavaram
> Sent: Wednesday, July 19, 2006 2:29 PM
> To: [hidden email]
> Subject: [protege-discussion] Re: I want some solid real
> world examples!
>
> Alan,
>  Let's say that i have an owl description of the ontology for
> Neuron (cell structures in brain that processes information).
> Now what? how does this owl file help me? according to my
> understanding, i know that having an owl file is like
> specifying the semantics of Neuron concept.
> Tell me what's the first use that comes to your mind seeing
> an owl file. you can probably use ur own example.
> -Thanks
> Sri
>
> On 7/18/06, Alan March <[hidden email]> wrote:
> >
> >
> > Yep. In our case its SQL Server. What I meant by "RDF" is
> that we use
> > a table which contains the info that would otherwise be
> contained in
> > the RDF/OWL File. Sorry, I loosely called it "RDF table" but that's
> > just a local designation we use to refer to that table (the "rdf"
> > table"...granted: it does no service to proper understanding of the
> > concept...true reification in the literal sense of the
> term). You can
> > actually call it any name in the JDBC implementation of Protégé.
> >
> > Regarding the use of "list": An adjacency list is a manner
> in which to
> > represent a graph structure and it can be implemented in a
> database table.
> > It's a classical implmentation of graph structures. Joe Celko has
> > written extensively on that, using ANSI-SQL.
> >
> > Finally, there is a lot of literature on how to map (that is,
> > represent) RDF structures to a RDBMS. The authors of the
> JDBC backend
> > made it simple and I actually think they did it very well. Its
> > supposed inefficiency can be dealt
> >
> > A word of advice: if you use SQL server as a backend, don't use
> > Microsoft JDBC driver. I think there's an issue regarding the Java
> > runtime, but it simply hangs when you try to use a reasoner, with a
> > message regarding "column 6" which I frankly did not
> research into. I
> > just dumped the driver.
> >
> > Regards,
> >
> > Alan
> >
> > > -----Original Message-----
> > > From: [hidden email]
> > > [mailto:[hidden email]] On Behalf Of
> > > sridevi polavaram
> > > Sent: Tuesday, July 18, 2006 3:23 PM
> > > To: [hidden email]
> > > Subject: [protege-discussion] Re: I want some solid real world
> > > examples!
> > >
> > > Alan,
> > > when you say database backend, i am assuming it is some
> RDBMS like
> > > mysql, oracle etc., but you are referring it as RDF table
> and a list
> > > structure! so what kind of DB is this! can u maintain owl or RDF
> > > DBs!
> > > what is the size limit of these DBs? and how exactly are you
> > > achieving data storage and retrieval methods!
> > > -regards,
> > > Sri.
> > >
> > > On 7/18/06, Alan March <[hidden email]> wrote:
> > > >
> > > >
> > > > Well, one of the first problems we run into is OWL itself.
> > > Although it
> > > > is very powefull in expressivity, we believe it is not
> > > efficient as a
> > > > storage and retrieval repository (indeed, it was probably
> > > not designed
> > > > with efficiency in mind, and in that sense it reminds
> me of my old
> > > > Prolog days).
> > > > So we used the JDBC database backend. The database backend
> > > is not that
> > > > efficient in itself (and, again, its creators mention in the
> > > > documentation that efficiency was not one of their
> > > requirements) but
> > > > it is extremely elastic. Stated simply, the DB backend is a
> > > (sort of)
> > > > adjacency list with the particularly interesting
> > > characteristic that
> > > > the list is double linked.
> > > > From what I gather, one important difference with the pure RDF
> > > > approach is that nodes are loaded on demand, and there
> is a slight
> > > > performance issue there, but then, it allows for managing large
> > > > ontologies. So all design is managed using Protégé as
> the front-end.
> > > >
> > > > The advantage of the database backend is that it makes all the
> > > > elements of the ontology easily accesible for other
> > > applications when
> > > > these applications aren't involved with the design task itself.
> > > > Example: the multiple classifications built by reasoners such as
> > > > Fact++ may be accessed from web pages and applications using
> > > > Fact++ simple
> > > > approaches such as treeviews and the like, according to user's
> > > > different classification requirements (a CFO is interested
> > > if a cost
> > > > based classification of procedures, whilst a chief medical
> > > officer is
> > > > interested in a body systems classification of procedures).
> > > > Another
> > > > example: extracting dimensions for a warehouse. And of course
> > > > there are more.
> > > >
> > > > Most of these tasks are carried out by writing queries to the
> > > > database. We use stored procedures, which are included
> in the same
> > > > database  that contains the main "RDF" table. The pattern
> > > > generally used is a self linked table for accessing the
> adjacency
> > > > list schema
> > > >
> > > > Don't know how much "state of the art" this is, but it
> works for us.
> > > >
> > > > Regards,
> > > >
> > > > AM
> > > >
> > > > > -----Original Message-----
> > > > > From: [hidden email]
> > > > > [mailto:[hidden email]]
> On Behalf Of
> > > > > sridevi polavaram
> > > > > Sent: Tuesday, July 18, 2006 9:35 AM
> > > > > To: [hidden email]
> > > > > Subject: [protege-discussion] Re: I want some solid
> real world
> > > > > examples!
> > > > >
> > > > > Alan,
> > > > >  Thankyou very much for explaining things so
> patiently. So, how
> > > > > exactly are you using the ontology that you have tailored
> > > for your
> > > > > purpose at the application level? are you mapping the
> ontology
> > > > > properties as attributes at the database-level?
> > > > > if so, what are the other kinds of mapping that u do?
> > > > > are u using any logic based programming languages? How is
> > > the rich
> > > > > annotation that's inherent in ontologies is helping
> you at the
> > > > > application level, apart from the fact that it serves as a
> > > > > reinforcement knowldge base for humans. how are you
> > > extracting and
> > > > > storing information using ontologies?  can you say that
> > > ontology is
> > > > > crucial for your application and that it's not
> replacable by any
> > > > > other design methodologies.
> > > > > I am sorry but my mind is full of questions...just can't stop
> > > > > the flow..i am reading all the related literature that's
> > > > > available (white papers, tools, theroies, real examples,
> > > > > technology
> > > etc.,) but
> > > > > none of them help me converge my understanding.
> > > > > -regards,
> > > > > Sri
> > > > >
> > > > > On 7/17/06, Alan March <[hidden email]> wrote:
> > > > > >
> > > > > >
> > > > > > > > Wow SNOMED and UMLS are supposingly success stories for
> > > > > > > > ontology
> > > > > > > applications. and they failed to mee your requirements!!
> > > > > > > then the whole point of "sharing knowldge" seems
> skeptical.
> > > > > > > So are u designing ontologies from scratch for your HMO
> > > > > application!
> > > > > > > everyone is apparently having there own ontology
> > > > > >
> > > > > > The trouble we faced with SNOMED and UMLS is they
> > > appeared to us
> > > > > > as incomplete as far as attributes (relations) are
> > > concerned: too
> > > > > > few, failing to capture the richness of possible relations.
> > > > > > And then of course, the mistakes found in them, which Barry
> > > Smith and
> > > > > others have
> > > > > > mentioned in their papers. Also, much of the terms we had
> > > > > to use were
> > > > > > highly local and neither SNOMED nor UMLS offered a clear
> > > > > > method for extensions. Although SNOMED has
> something on those
> > > > > > lines,
> > > > > the problem
> > > > > > of its "ontological incompleteness"
> > > > > > and licensing questions represented a drawback for us. But
> > > > > then again:
> > > > > > both
> > > > > > SNOMED and UMLS have been extremely usefull to us
> as a sort of
> > > > > > knowledge repositories, but we just could not use them
> > > > > "out-of-the-box".
> > > > > >
> > > > > > Actually, we didn't start from scratch: we reutilized
> > > much of the
> > > > > > publically available Galen files, and modelled our ontology
> > > > > following
> > > > > > an upper level ontology. For anatomy we used Digital
> > > Anatomist as
> > > > > > published in UMLS. We tried to use as much as possible from
> > > > > > several ontologies we found to be suitable to our purposes.
> > > > > > But
> > > > > yes, we had to
> > > > > > reinvent the wheel in some circumstances. In any case, I am
> > > > > not quite
> > > > > > sure if we will ever manage to arrive at a universally
> > > > > > accepted ontology. Probably at the upper and middle levels
> > > > > > this will be achieved, but I doubt it will work for "bottom"
> > > > > > levels.
> > > > > >
> > > > > > Regarding success stories: well, I guess SNOMED is a
> > > > > success story in
> > > > > > a certain sense, but I am not quite sure if it can be
> > > > > called that way
> > > > > > as far as ontological principles are concerned. Not to deny
> > > > > its value:
> > > > > > it is rich in content. Worth mentioning is the fact
> > > that we didn't
> > > > > > want to consider terminologies which weren't in the public
> > > > > domain or commercial in nature.
> > > > > > I
> > > > > > personally feel that medical terminologies such be
> built in a
> > > > > > collaborative manner, probable on the same lines as Linux
> > > > > was made (by
> > > > > > having different groups managing different parts of the
> > > domain).
> > > > > > In that sense, I feel that Rector's advice regarding the way
> > > > > ontologies
> > > > > > should be built (by
> > > > > > "untangling") allows for this. I am relly sorry that the
> > > > > Galen project
> > > > > > didn't continue. Looking back on it, I think it is
> > > still the best
> > > > > > thing we have seen in all these decades of discussion on
> > > > > terminologies
> > > > > > and how they should be built and maintained.
> > > > > >
> > > > > > >
> > > > > > > In that sense, I find that ontologies are "the" techique
> > > > > for design,
> > > > > > > > development and maintaince of so-called controlled
> > > > > terminologies.
> > > > > > >
> > > > > > >
> > > > > > > > yes, i would say, that even i am convinced that at
> > > the level
> > > > > > > > of
> > > > > > > vocabularies. like a dictionary sort of thing. the
> > > wikipedia for
> > > > > > > encyclopedia also uses ontologies!!
> > > > > > > Thanks a lot for giving me some idea about the real workd
> > > > > scenario.
> > > > > > > Best,
> > > > > > > -Sri
> > > > > > >
> > > > > > > OF course,
> > > > > > > > in the process of using ontologies one quickly realizes
> > > > > > > > the
> > > > > > > fact that
> > > > > > > > they could probably be usefull for analysis and design
> > > > > of complex
> > > > > > > > systems, and I personally feel that they could
> > > > > ultimately replace
> > > > > > > > other tools such as UML and the like. This, of
> course (and
> > > > > > > at least as
> > > > > > > > far as my comprehension goes), still requires much work
> > > > > regarding
> > > > > > > > methodology and probably software tools. But I feel that
> > > > > > > there is more
> > > > > > > > to ontologies than what meets the eye of the casual
> > > > > user. Example:
> > > > > > > > I've been following developments in HL7 v3 and am every
> > > > > day more
> > > > > > > > convinced that v3 should be redesigned using readily
> > > > > > > available tools
> > > > > > > > such as Protégé (I don't think it shoud be
> thrown away as
> > > > > > > others have
> > > > > > > > suggestes, as the RIM contains a lot of useful domain
> > > > > > > information, but
> > > > > > > > the methodolgy user does not seem adequate).
> > > > > > > >
> > > > > > > > So, I personally feel that ontologies and Protégé-type
> > > > > tools could
> > > > > > > > turn out to be a small revolution in systems
> > > design. Should my
> > > > > > > > idea prove right, we are now in a sort of
> pre-paradigmatic
> > > > > > > situation as far
> > > > > > > > as application of ontologies is concerened.
> > > > > > > >
> > > > > > > > Best regards
> > > > > > > >
> > > > > > > > > -----Original Message-----
> > > > > > > > > From: [hidden email]
> > > > > > > > > [mailto:[hidden email]]
> > > > > On Behalf Of
> > > > > > > > > sridevi polavaram
> > > > > > > > > Sent: Monday, July 17, 2006 12:52 PM
> > > > > > > > > To: [hidden email]
> > > > > > > > > Subject: [protege-discussion] I want some solid real
> > > > > > > world examples!
> > > > > > > > >
> > > > > > > > > Hello All,
> > > > > > > > >  Please help me find an answer to my questions. I
> > > > > suppose some
> > > > > > > > > of you might also have the same questions. so let's
> > > > > brainstorm.
> > > > > > > > > I totally understand what ontology is
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > >
> > >
> --------------------------------------------------------------------
> > > > > > > --
> > > > > > > > --- To unsubscribe go to
> > > > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> ------------------------------------------------------------
> > > > > > > --
> > > > > > > -----------
> > > > > > > To unsubscribe go to
> > > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > >
> --------------------------------------------------------------------
> > > > > --
> > > > > > --- To unsubscribe go to
> > > > > > http://protege.stanford.edu/community/subscribe.html
> > > > > >
> > > > > >
> > > > >
> > > > > --------------------------------------------------------------
> > > > > -----------
> > > > > To unsubscribe go to
> > > > > http://protege.stanford.edu/community/subscribe.html
> > > > >
> > > > >
> > > >
> > > >
> > >
> --------------------------------------------------------------------
> > > --
> > > > --- To unsubscribe go to
> > > > http://protege.stanford.edu/community/subscribe.html
> > > >
> > > >
> > >
> > > --------------------------------------------------------------
> > > -----------
> > > To unsubscribe go to
> > > http://protege.stanford.edu/community/subscribe.html
> > >
> >
> >
> ----------------------------------------------------------------------
> > --- To unsubscribe go to
> > http://protege.stanford.edu/community/subscribe.html
> >
> >
>
> --------------------------------------------------------------
> -----------
> To unsubscribe go to
> http://protege.stanford.edu/community/subscribe.html
>

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html


-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html


-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

Reply | Threaded
Open this post in threaded view
|

Re: Use of ontologies in software development

Leslie Barrett
In reply to this post by NEHEMIAH MAVETERA
 

Hello -

Although it ihas fallen out of favor recently, WordNet was called at runtime
for query expansion in various question-answering systems.

Leslie Barrett


----------------------------------------------------------------------------
From:  "NEHEMIAH MAVETERA" <[hidden email]>
Reply-To:  [hidden email]
To:  <[hidden email]>, <[hidden email]>
Subject:  [protege-discussion] Use of ontologies in software development
Date:  Thu, 20 Jul 2006 16:07:20 +0200
>
>Dear All
>
>Is there anyone who has used ontologies in software development. I mean at
run time.  Please furnish me with any projects that you have done in this
regard.
>
>My problems are that, as a systems developer, analyst you name it, what
should I look for in a software development project that can be enhanced
using ontologies.
>
>A typical example is " Use of ontologies in software development for agent
mediated electronic marktes" . How do I make use of ontologies at run time?
>
>Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA;
MSAICSIT; (SA))
>Senior Lecturer- Information Systems
>North West University
>www.uniwest.ac.za
>Tel: +27183892143
>Fax: +27183892090
>
>Can we isolate reason, and is it, so regarded, an independent source of
concepts and judgements which spring from it alone? From  I. Kant, Critique
of Pure Reason, A305-B362

>>>>[hidden email] 07/20/06 2:13 AM >>>
>
>Well, probably someone else could answer that better than me, but I feel
>there are some comments I would venture to make.
>
>An ontology for a neuron should imply an ontology for a cell. So, if
>properly designed, should contain a potential for a complete cellular and
>subcellular ontology. This is, to the best of my knowledge, what has been
>done in the Formal Model of Anatomy. In the context of our project, which I
>mentioned earlier, we have used Digital Anatomist for modelling anatomy as
>it pertains to surgical procedures. So, although I am streching the
concept,
>in a certain sense an ontology for cells is part of our project. We have
not
>as yet started with the ontology of laboratory procedures, but when we
reach
>that point I am sure we will be using the cell ontology which you
indirectly

>mention, proably drawn form Digital Anatomist.
>
>Hope my comments have helped.
>
>>-----Original Message-----
>>From: [hidden email]
>>[mailto:[hidden email]] On Behalf
>>Of sridevi polavaram
>>Sent: Wednesday, July 19, 2006 2:29 PM
>>To: [hidden email]
>>Subject: [protege-discussion] Re: I want some solid real
>>world examples!
>>
>>Alan,
>>  Let's say that i have an owl description of the ontology for
>>Neuron (cell structures in brain that processes information).
>>Now what? how does this owl file help me? according to my
>>understanding, i know that having an owl file is like
>>specifying the semantics of Neuron concept.
>>Tell me what's the first use that comes to your mind seeing
>>an owl file. you can probably use ur own example.
>>-Thanks
>>Sri
>>
>>On 7/18/06, Alan March <[hidden email]> wrote:
>>>
>>>
>>>Yep. In our case its SQL Server. What I meant by "RDF" is
>>that we use
>>>a table which contains the info that would otherwise be
>>contained in
>>>the RDF/OWL File. Sorry, I loosely called it "RDF table" but that's
>>>just a local designation we use to refer to that table (the "rdf"
>>>table"...granted: it does no service to proper understanding of the
>>>concept...true reification in the literal sense of the
>>term). You can
>>>actually call it any name in the JDBC implementation of Protégé.
>>>
>>>Regarding the use of "list": An adjacency list is a manner
>>in which to
>>>represent a graph structure and it can be implemented in a
>>database table.
>>>It's a classical implmentation of graph structures. Joe Celko has
>>>written extensively on that, using ANSI-SQL.
>>>
>>>Finally, there is a lot of literature on how to map (that is,
>>>represent) RDF structures to a RDBMS. The authors of the
>>JDBC backend
>>>made it simple and I actually think they did it very well. Its
>>>supposed inefficiency can be dealt
>>>
>>>A word of advice: if you use SQL server as a backend, don't use
>>>Microsoft JDBC driver. I think there's an issue regarding the Java
>>>runtime, but it simply hangs when you try to use a reasoner, with a
>>>message regarding "column 6" which I frankly did not
>>research into. I
>>>just dumped the driver.
>>>
>>>Regards,
>>>
>>>Alan
>>>
>>>>-----Original Message-----
>>>>From: [hidden email]
>>>>[mailto:[hidden email]] On Behalf Of
>>>>sridevi polavaram
>>>>Sent: Tuesday, July 18, 2006 3:23 PM
>>>>To: [hidden email]
>>>>Subject: [protege-discussion] Re: I want some solid real world
>>>>examples!
>>>>
>>>>Alan,
>>>>when you say database backend, i am assuming it is some
>>RDBMS like
>>>>mysql, oracle etc., but you are referring it as RDF table
>>and a list
>>>>structure! so what kind of DB is this! can u maintain owl or RDF
>>>>DBs!
>>>>what is the size limit of these DBs? and how exactly are you
>>>>achieving data storage and retrieval methods!
>>>>-regards,
>>>>Sri.
>>>>
>>>>On 7/18/06, Alan March <[hidden email]> wrote:
>>>>>
>>>>>
>>>>>Well, one of the first problems we run into is OWL itself.
>>>>Although it
>>>>>is very powefull in expressivity, we believe it is not
>>>>efficient as a
>>>>>storage and retrieval repository (indeed, it was probably
>>>>not designed
>>>>>with efficiency in mind, and in that sense it reminds
>>me of my old
>>>>>Prolog days).
>>>>>So we used the JDBC database backend. The database backend
>>>>is not that
>>>>>efficient in itself (and, again, its creators mention in the
>>>>>documentation that efficiency was not one of their
>>>>requirements) but
>>>>>it is extremely elastic. Stated simply, the DB backend is a
>>>>(sort of)
>>>>>adjacency list with the particularly interesting
>>>>characteristic that
>>>>>the list is double linked.
>>>>>From what I gather, one important difference with the pure RDF
>>>>>approach is that nodes are loaded on demand, and there
>>is a slight
>>>>>performance issue there, but then, it allows for managing large
>>>>>ontologies. So all design is managed using Protégé as
>>the front-end.
>>>>>
>>>>>The advantage of the database backend is that it makes all the
>>>>>elements of the ontology easily accesible for other
>>>>applications when
>>>>>these applications aren't involved with the design task itself.
>>>>>Example: the multiple classifications built by reasoners such as
>>>>>Fact++ may be accessed from web pages and applications using
>>>>>Fact++ simple
>>>>>approaches such as treeviews and the like, according to user's
>>>>>different classification requirements (a CFO is interested
>>>>if a cost
>>>>>based classification of procedures, whilst a chief medical
>>>>officer is
>>>>>interested in a body systems classification of procedures).
>>>>>Another
>>>>>example: extracting dimensions for a warehouse. And of course
>>>>>there are more.
>>>>>
>>>>>Most of these tasks are carried out by writing queries to the
>>>>>database. We use stored procedures, which are included
>>in the same
>>>>>database  that contains the main "RDF" table. The pattern
>>>>>generally used is a self linked table for accessing the
>>adjacency
>>>>>list schema
>>>>>
>>>>>Don't know how much "state of the art" this is, but it
>>works for us.
>>>>>
>>>>>Regards,
>>>>>
>>>>>AM
>>>>>
>>>>>>-----Original Message-----
>>>>>>From: [hidden email]
>>>>>>[mailto:[hidden email]]
>>On Behalf Of
>>>>>>sridevi polavaram
>>>>>>Sent: Tuesday, July 18, 2006 9:35 AM
>>>>>>To: [hidden email]
>>>>>>Subject: [protege-discussion] Re: I want some solid
>>real world
>>>>>>examples!
>>>>>>
>>>>>>Alan,
>>>>>>  Thankyou very much for explaining things so
>>patiently. So, how
>>>>>>exactly are you using the ontology that you have tailored
>>>>for your
>>>>>>purpose at the application level? are you mapping the
>>ontology
>>>>>>properties as attributes at the database-level?
>>>>>>if so, what are the other kinds of mapping that u do?
>>>>>>are u using any logic based programming languages? How is
>>>>the rich
>>>>>>annotation that's inherent in ontologies is helping
>>you at the
>>>>>>application level, apart from the fact that it serves as a
>>>>>>reinforcement knowldge base for humans. how are you
>>>>extracting and
>>>>>>storing information using ontologies?  can you say that
>>>>ontology is
>>>>>>crucial for your application and that it's not
>>replacable by any
>>>>>>other design methodologies.
>>>>>>I am sorry but my mind is full of questions...just can't stop
>>>>>>the flow..i am reading all the related literature that's
>>>>>>available (white papers, tools, theroies, real examples,
>>>>>>technology
>>>>etc.,) but
>>>>>>none of them help me converge my understanding.
>>>>>>-regards,
>>>>>>Sri
>>>>>>
>>>>>>On 7/17/06, Alan March <[hidden email]> wrote:
>>>>>>>
>>>>>>>
>>>>>>>>>Wow SNOMED and UMLS are supposingly success stories for
>>>>>>>>>ontology
>>>>>>>>applications. and they failed to mee your requirements!!
>>>>>>>>then the whole point of "sharing knowldge" seems
>>skeptical.
>>>>>>>>So are u designing ontologies from scratch for your HMO
>>>>>>application!
>>>>>>>>everyone is apparently having there own ontology
>>>>>>>
>>>>>>>The trouble we faced with SNOMED and UMLS is they
>>>>appeared to us
>>>>>>>as incomplete as far as attributes (relations) are
>>>>concerned: too
>>>>>>>few, failing to capture the richness of possible relations.
>>>>>>>And then of course, the mistakes found in them, which Barry
>>>>Smith and
>>>>>>others have
>>>>>>>mentioned in their papers. Also, much of the terms we had
>>>>>>to use were
>>>>>>>highly local and neither SNOMED nor UMLS offered a clear
>>>>>>>method for extensions. Although SNOMED has
>>something on those
>>>>>>>lines,
>>>>>>the problem
>>>>>>>of its "ontological incompleteness"
>>>>>>>and licensing questions represented a drawback for us. But
>>>>>>then again:
>>>>>>>both
>>>>>>>SNOMED and UMLS have been extremely usefull to us
>>as a sort of
>>>>>>>knowledge repositories, but we just could not use them
>>>>>>"out-of-the-box".
>>>>>>>
>>>>>>>Actually, we didn't start from scratch: we reutilized
>>>>much of the
>>>>>>>publically available Galen files, and modelled our ontology
>>>>>>following
>>>>>>>an upper level ontology. For anatomy we used Digital
>>>>Anatomist as
>>>>>>>published in UMLS. We tried to use as much as possible from
>>>>>>>several ontologies we found to be suitable to our purposes.
>>>>>>>But
>>>>>>yes, we had to
>>>>>>>reinvent the wheel in some circumstances. In any case, I am
>>>>>>not quite
>>>>>>>sure if we will ever manage to arrive at a universally
>>>>>>>accepted ontology. Probably at the upper and middle levels
>>>>>>>this will be achieved, but I doubt it will work for "bottom"
>>>>>>>levels.
>>>>>>>
>>>>>>>Regarding success stories: well, I guess SNOMED is a
>>>>>>success story in
>>>>>>>a certain sense, but I am not quite sure if it can be
>>>>>>called that way
>>>>>>>as far as ontological principles are concerned. Not to deny
>>>>>>its value:
>>>>>>>it is rich in content. Worth mentioning is the fact
>>>>that we didn't
>>>>>>>want to consider terminologies which weren't in the public
>>>>>>domain or commercial in nature.
>>>>>>>I
>>>>>>>personally feel that medical terminologies such be
>>built in a
>>>>>>>collaborative manner, probable on the same lines as Linux
>>>>>>was made (by
>>>>>>>having different groups managing different parts of the
>>>>domain).
>>>>>>>In that sense, I feel that Rector's advice regarding the way
>>>>>>ontologies
>>>>>>>should be built (by
>>>>>>>"untangling") allows for this. I am relly sorry that the
>>>>>>Galen project
>>>>>>>didn't continue. Looking back on it, I think it is
>>>>still the best
>>>>>>>thing we have seen in all these decades of discussion on
>>>>>>terminologies
>>>>>>>and how they should be built and maintained.
>>>>>>>
>>>>>>>>
>>>>>>>>In that sense, I find that ontologies are "the" techique
>>>>>>for design,
>>>>>>>>>development and maintaince of so-called controlled
>>>>>>terminologies.
>>>>>>>>
>>>>>>>>
>>>>>>>>>yes, i would say, that even i am convinced that at
>>>>the level
>>>>>>>>>of
>>>>>>>>vocabularies. like a dictionary sort of thing. the
>>>>wikipedia for
>>>>>>>>encyclopedia also uses ontologies!!
>>>>>>>>Thanks a lot for giving me some idea about the real workd
>>>>>>scenario.
>>>>>>>>Best,
>>>>>>>>-Sri
>>>>>>>>
>>>>>>>>OF course,
>>>>>>>>>in the process of using ontologies one quickly realizes
>>>>>>>>>the
>>>>>>>>fact that
>>>>>>>>>they could probably be usefull for analysis and design
>>>>>>of complex
>>>>>>>>>systems, and I personally feel that they could
>>>>>>ultimately replace
>>>>>>>>>other tools such as UML and the like. This, of
>>course (and
>>>>>>>>at least as
>>>>>>>>>far as my comprehension goes), still requires much work
>>>>>>regarding
>>>>>>>>>methodology and probably software tools. But I feel that
>>>>>>>>there is more
>>>>>>>>>to ontologies than what meets the eye of the casual
>>>>>>user. Example:
>>>>>>>>>I've been following developments in HL7 v3 and am every
>>>>>>day more
>>>>>>>>>convinced that v3 should be redesigned using readily
>>>>>>>>available tools
>>>>>>>>>such as Protégé (I don't think it shoud be
>>thrown away as
>>>>>>>>others have
>>>>>>>>>suggestes, as the RIM contains a lot of useful domain
>>>>>>>>information, but
>>>>>>>>>the methodolgy user does not seem adequate).
>>>>>>>>>
>>>>>>>>>So, I personally feel that ontologies and Protégé-type
>>>>>>tools could
>>>>>>>>>turn out to be a small revolution in systems
>>>>design. Should my
>>>>>>>>>idea prove right, we are now in a sort of
>>pre-paradigmatic
>>>>>>>>situation as far
>>>>>>>>>as application of ontologies is concerened.
>>>>>>>>>
>>>>>>>>>Best regards
>>>>>>>>>
>>>>>>>>>>-----Original Message-----
>>>>>>>>>>From: [hidden email]
>>>>>>>>>>[mailto:[hidden email]]
>>>>>>On Behalf Of
>>>>>>>>>>sridevi polavaram
>>>>>>>>>>Sent: Monday, July 17, 2006 12:52 PM
>>>>>>>>>>To: [hidden email]
>>>>>>>>>>Subject: [protege-discussion] I want some solid real
>>>>>>>>world examples!
>>>>>>>>>>
>>>>>>>>>>Hello All,
>>>>>>>>>>  Please help me find an answer to my questions. I
>>>>>>suppose some
>>>>>>>>>>of you might also have the same questions. so let's
>>>>>>brainstorm.
>>>>>>>>>>I totally understand what ontology is
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>
>>>>
>>--------------------------------------------------------------------
>>>>>>>>--
>>>>>>>>>--- To unsubscribe go to
>>>>>>>>>http://protege.stanford.edu/community/subscribe.html
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>------------------------------------------------------------
>>>>>>>>--
>>>>>>>>-----------
>>>>>>>>To unsubscribe go to
>>>>>>>>http://protege.stanford.edu/community/subscribe.html
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>
>>--------------------------------------------------------------------
>>>>>>--
>>>>>>>--- To unsubscribe go to
>>>>>>>http://protege.stanford.edu/community/subscribe.html
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>--------------------------------------------------------------
>>>>>>-----------
>>>>>>To unsubscribe go to
>>>>>>http://protege.stanford.edu/community/subscribe.html
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>--------------------------------------------------------------------
>>>>--
>>>>>--- To unsubscribe go to
>>>>>http://protege.stanford.edu/community/subscribe.html
>>>>>
>>>>>
>>>>
>>>>--------------------------------------------------------------
>>>>-----------
>>>>To unsubscribe go to
>>>>http://protege.stanford.edu/community/subscribe.html
>>>>
>>>
>>>
>>----------------------------------------------------------------------
>>>--- To unsubscribe go to
>>>http://protege.stanford.edu/community/subscribe.html
>>>
>>>
>>
>>--------------------------------------------------------------
>>-----------
>>To unsubscribe go to
>>http://protege.stanford.edu/community/subscribe.html
>>
>
>-------------------------------------------------------------------------
>To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
>
>
>-------------------------------------------------------------------------
>To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
>

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

Reply | Threaded
Open this post in threaded view
|

Re: Use of ontologies in software development

Nuno Seco

Hello - Not sure if this relevant, but here goes anyway...

There is a project called ReBuilder
(http://ailab.dei.uc.pt/view_project.php?id_p=5 ) that does make use of
ontologies for software design. The project started out using WordNet (a
lexical ontology) but is now moving towards domain specific ontologies in
its second version.

--
Nuno Seco

> -----Original Message-----
> From: [hidden email] [mailto:protege-
> [hidden email]] On Behalf Of Leslie Barrett
> Sent: quinta-feira, 20 de Julho de 2006 18:30
> To: [hidden email]
> Subject: [protege-discussion] Re: Use of ontologies in software
> development
>
>
>
> Hello -
>
> Although it ihas fallen out of favor recently, WordNet was called at
> runtime
> for query expansion in various question-answering systems.
>
> Leslie Barrett
>
>
> --------------------------------------------------------------------------
> --
> From:  "NEHEMIAH MAVETERA" <[hidden email]>
> Reply-To:  [hidden email]
> To:  <[hidden email]>, <[hidden email]>
> Subject:  [protege-discussion] Use of ontologies in software development
> Date:  Thu, 20 Jul 2006 16:07:20 +0200
> >
> >Dear All
> >
> >Is there anyone who has used ontologies in software development. I mean
> at
> run time.  Please furnish me with any projects that you have done in this
> regard.
> >
> >My problems are that, as a systems developer, analyst you name it, what
> should I look for in a software development project that can be enhanced
> using ontologies.
> >
> >A typical example is " Use of ontologies in software development for
> agent
> mediated electronic marktes" . How do I make use of ontologies at run
> time?
> >
> >Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA;
> MSAICSIT; (SA))
> >Senior Lecturer- Information Systems
> >North West University
> >www.uniwest.ac.za
> >Tel: +27183892143
> >Fax: +27183892090
> >
> >Can we isolate reason, and is it, so regarded, an independent source of
> concepts and judgements which spring from it alone? From  I. Kant,
> Critique
> of Pure Reason, A305-B362
> >>>>[hidden email] 07/20/06 2:13 AM >>>
> >
> >Well, probably someone else could answer that better than me, but I feel
> >there are some comments I would venture to make.
> >
> >An ontology for a neuron should imply an ontology for a cell. So, if
> >properly designed, should contain a potential for a complete cellular and
> >subcellular ontology. This is, to the best of my knowledge, what has been
> >done in the Formal Model of Anatomy. In the context of our project, which
> I
> >mentioned earlier, we have used Digital Anatomist for modelling anatomy
> as
> >it pertains to surgical procedures. So, although I am streching the
> concept,
> >in a certain sense an ontology for cells is part of our project. We have
> not
> >as yet started with the ontology of laboratory procedures, but when we
> reach
> >that point I am sure we will be using the cell ontology which you
> indirectly
> >mention, proably drawn form Digital Anatomist.
> >
> >Hope my comments have helped.
> >
> >>-----Original Message-----
> >>From: [hidden email]
> >>[mailto:[hidden email]] On Behalf
> >>Of sridevi polavaram
> >>Sent: Wednesday, July 19, 2006 2:29 PM
> >>To: [hidden email]
> >>Subject: [protege-discussion] Re: I want some solid real
> >>world examples!
> >>
> >>Alan,
> >>  Let's say that i have an owl description of the ontology for
> >>Neuron (cell structures in brain that processes information).
> >>Now what? how does this owl file help me? according to my
> >>understanding, i know that having an owl file is like
> >>specifying the semantics of Neuron concept.
> >>Tell me what's the first use that comes to your mind seeing
> >>an owl file. you can probably use ur own example.
> >>-Thanks
> >>Sri
> >>
> >>On 7/18/06, Alan March <[hidden email]> wrote:
> >>>
> >>>
> >>>Yep. In our case its SQL Server. What I meant by "RDF" is
> >>that we use
> >>>a table which contains the info that would otherwise be
> >>contained in
> >>>the RDF/OWL File. Sorry, I loosely called it "RDF table" but that's
> >>>just a local designation we use to refer to that table (the "rdf"
> >>>table"...granted: it does no service to proper understanding of the
> >>>concept...true reification in the literal sense of the
> >>term). You can
> >>>actually call it any name in the JDBC implementation of Protégé.
> >>>
> >>>Regarding the use of "list": An adjacency list is a manner
> >>in which to
> >>>represent a graph structure and it can be implemented in a
> >>database table.
> >>>It's a classical implmentation of graph structures. Joe Celko has
> >>>written extensively on that, using ANSI-SQL.
> >>>
> >>>Finally, there is a lot of literature on how to map (that is,
> >>>represent) RDF structures to a RDBMS. The authors of the
> >>JDBC backend
> >>>made it simple and I actually think they did it very well. Its
> >>>supposed inefficiency can be dealt
> >>>
> >>>A word of advice: if you use SQL server as a backend, don't use
> >>>Microsoft JDBC driver. I think there's an issue regarding the Java
> >>>runtime, but it simply hangs when you try to use a reasoner, with a
> >>>message regarding "column 6" which I frankly did not
> >>research into. I
> >>>just dumped the driver.
> >>>
> >>>Regards,
> >>>
> >>>Alan
> >>>
> >>>>-----Original Message-----
> >>>>From: [hidden email]
> >>>>[mailto:[hidden email]] On Behalf Of
> >>>>sridevi polavaram
> >>>>Sent: Tuesday, July 18, 2006 3:23 PM
> >>>>To: [hidden email]
> >>>>Subject: [protege-discussion] Re: I want some solid real world
> >>>>examples!
> >>>>
> >>>>Alan,
> >>>>when you say database backend, i am assuming it is some
> >>RDBMS like
> >>>>mysql, oracle etc., but you are referring it as RDF table
> >>and a list
> >>>>structure! so what kind of DB is this! can u maintain owl or RDF
> >>>>DBs!
> >>>>what is the size limit of these DBs? and how exactly are you
> >>>>achieving data storage and retrieval methods!
> >>>>-regards,
> >>>>Sri.
> >>>>
> >>>>On 7/18/06, Alan March <[hidden email]> wrote:
> >>>>>
> >>>>>
> >>>>>Well, one of the first problems we run into is OWL itself.
> >>>>Although it
> >>>>>is very powefull in expressivity, we believe it is not
> >>>>efficient as a
> >>>>>storage and retrieval repository (indeed, it was probably
> >>>>not designed
> >>>>>with efficiency in mind, and in that sense it reminds
> >>me of my old
> >>>>>Prolog days).
> >>>>>So we used the JDBC database backend. The database backend
> >>>>is not that
> >>>>>efficient in itself (and, again, its creators mention in the
> >>>>>documentation that efficiency was not one of their
> >>>>requirements) but
> >>>>>it is extremely elastic. Stated simply, the DB backend is a
> >>>>(sort of)
> >>>>>adjacency list with the particularly interesting
> >>>>characteristic that
> >>>>>the list is double linked.
> >>>>>From what I gather, one important difference with the pure RDF
> >>>>>approach is that nodes are loaded on demand, and there
> >>is a slight
> >>>>>performance issue there, but then, it allows for managing large
> >>>>>ontologies. So all design is managed using Protégé as
> >>the front-end.
> >>>>>
> >>>>>The advantage of the database backend is that it makes all the
> >>>>>elements of the ontology easily accesible for other
> >>>>applications when
> >>>>>these applications aren't involved with the design task itself.
> >>>>>Example: the multiple classifications built by reasoners such as
> >>>>>Fact++ may be accessed from web pages and applications using
> >>>>>Fact++ simple
> >>>>>approaches such as treeviews and the like, according to user's
> >>>>>different classification requirements (a CFO is interested
> >>>>if a cost
> >>>>>based classification of procedures, whilst a chief medical
> >>>>officer is
> >>>>>interested in a body systems classification of procedures).
> >>>>>Another
> >>>>>example: extracting dimensions for a warehouse. And of course
> >>>>>there are more.
> >>>>>
> >>>>>Most of these tasks are carried out by writing queries to the
> >>>>>database. We use stored procedures, which are included
> >>in the same
> >>>>>database  that contains the main "RDF" table. The pattern
> >>>>>generally used is a self linked table for accessing the
> >>adjacency
> >>>>>list schema
> >>>>>
> >>>>>Don't know how much "state of the art" this is, but it
> >>works for us.
> >>>>>
> >>>>>Regards,
> >>>>>
> >>>>>AM
> >>>>>
> >>>>>>-----Original Message-----
> >>>>>>From: [hidden email]
> >>>>>>[mailto:[hidden email]]
> >>On Behalf Of
> >>>>>>sridevi polavaram
> >>>>>>Sent: Tuesday, July 18, 2006 9:35 AM
> >>>>>>To: [hidden email]
> >>>>>>Subject: [protege-discussion] Re: I want some solid
> >>real world
> >>>>>>examples!
> >>>>>>
> >>>>>>Alan,
> >>>>>>  Thankyou very much for explaining things so
> >>patiently. So, how
> >>>>>>exactly are you using the ontology that you have tailored
> >>>>for your
> >>>>>>purpose at the application level? are you mapping the
> >>ontology
> >>>>>>properties as attributes at the database-level?
> >>>>>>if so, what are the other kinds of mapping that u do?
> >>>>>>are u using any logic based programming languages? How is
> >>>>the rich
> >>>>>>annotation that's inherent in ontologies is helping
> >>you at the
> >>>>>>application level, apart from the fact that it serves as a
> >>>>>>reinforcement knowldge base for humans. how are you
> >>>>extracting and
> >>>>>>storing information using ontologies?  can you say that
> >>>>ontology is
> >>>>>>crucial for your application and that it's not
> >>replacable by any
> >>>>>>other design methodologies.
> >>>>>>I am sorry but my mind is full of questions...just can't stop
> >>>>>>the flow..i am reading all the related literature that's
> >>>>>>available (white papers, tools, theroies, real examples,
> >>>>>>technology
> >>>>etc.,) but
> >>>>>>none of them help me converge my understanding.
> >>>>>>-regards,
> >>>>>>Sri
> >>>>>>
> >>>>>>On 7/17/06, Alan March <[hidden email]> wrote:
> >>>>>>>
> >>>>>>>
> >>>>>>>>>Wow SNOMED and UMLS are supposingly success stories for
> >>>>>>>>>ontology
> >>>>>>>>applications. and they failed to mee your requirements!!
> >>>>>>>>then the whole point of "sharing knowldge" seems
> >>skeptical.
> >>>>>>>>So are u designing ontologies from scratch for your HMO
> >>>>>>application!
> >>>>>>>>everyone is apparently having there own ontology
> >>>>>>>
> >>>>>>>The trouble we faced with SNOMED and UMLS is they
> >>>>appeared to us
> >>>>>>>as incomplete as far as attributes (relations) are
> >>>>concerned: too
> >>>>>>>few, failing to capture the richness of possible relations.
> >>>>>>>And then of course, the mistakes found in them, which Barry
> >>>>Smith and
> >>>>>>others have
> >>>>>>>mentioned in their papers. Also, much of the terms we had
> >>>>>>to use were
> >>>>>>>highly local and neither SNOMED nor UMLS offered a clear
> >>>>>>>method for extensions. Although SNOMED has
> >>something on those
> >>>>>>>lines,
> >>>>>>the problem
> >>>>>>>of its "ontological incompleteness"
> >>>>>>>and licensing questions represented a drawback for us. But
> >>>>>>then again:
> >>>>>>>both
> >>>>>>>SNOMED and UMLS have been extremely usefull to us
> >>as a sort of
> >>>>>>>knowledge repositories, but we just could not use them
> >>>>>>"out-of-the-box".
> >>>>>>>
> >>>>>>>Actually, we didn't start from scratch: we reutilized
> >>>>much of the
> >>>>>>>publically available Galen files, and modelled our ontology
> >>>>>>following
> >>>>>>>an upper level ontology. For anatomy we used Digital
> >>>>Anatomist as
> >>>>>>>published in UMLS. We tried to use as much as possible from
> >>>>>>>several ontologies we found to be suitable to our purposes.
> >>>>>>>But
> >>>>>>yes, we had to
> >>>>>>>reinvent the wheel in some circumstances. In any case, I am
> >>>>>>not quite
> >>>>>>>sure if we will ever manage to arrive at a universally
> >>>>>>>accepted ontology. Probably at the upper and middle levels
> >>>>>>>this will be achieved, but I doubt it will work for "bottom"
> >>>>>>>levels.
> >>>>>>>
> >>>>>>>Regarding success stories: well, I guess SNOMED is a
> >>>>>>success story in
> >>>>>>>a certain sense, but I am not quite sure if it can be
> >>>>>>called that way
> >>>>>>>as far as ontological principles are concerned. Not to deny
> >>>>>>its value:
> >>>>>>>it is rich in content. Worth mentioning is the fact
> >>>>that we didn't
> >>>>>>>want to consider terminologies which weren't in the public
> >>>>>>domain or commercial in nature.
> >>>>>>>I
> >>>>>>>personally feel that medical terminologies such be
> >>built in a
> >>>>>>>collaborative manner, probable on the same lines as Linux
> >>>>>>was made (by
> >>>>>>>having different groups managing different parts of the
> >>>>domain).
> >>>>>>>In that sense, I feel that Rector's advice regarding the way
> >>>>>>ontologies
> >>>>>>>should be built (by
> >>>>>>>"untangling") allows for this. I am relly sorry that the
> >>>>>>Galen project
> >>>>>>>didn't continue. Looking back on it, I think it is
> >>>>still the best
> >>>>>>>thing we have seen in all these decades of discussion on
> >>>>>>terminologies
> >>>>>>>and how they should be built and maintained.
> >>>>>>>
> >>>>>>>>
> >>>>>>>>In that sense, I find that ontologies are "the" techique
> >>>>>>for design,
> >>>>>>>>>development and maintaince of so-called controlled
> >>>>>>terminologies.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>yes, i would say, that even i am convinced that at
> >>>>the level
> >>>>>>>>>of
> >>>>>>>>vocabularies. like a dictionary sort of thing. the
> >>>>wikipedia for
> >>>>>>>>encyclopedia also uses ontologies!!
> >>>>>>>>Thanks a lot for giving me some idea about the real workd
> >>>>>>scenario.
> >>>>>>>>Best,
> >>>>>>>>-Sri
> >>>>>>>>
> >>>>>>>>OF course,
> >>>>>>>>>in the process of using ontologies one quickly realizes
> >>>>>>>>>the
> >>>>>>>>fact that
> >>>>>>>>>they could probably be usefull for analysis and design
> >>>>>>of complex
> >>>>>>>>>systems, and I personally feel that they could
> >>>>>>ultimately replace
> >>>>>>>>>other tools such as UML and the like. This, of
> >>course (and
> >>>>>>>>at least as
> >>>>>>>>>far as my comprehension goes), still requires much work
> >>>>>>regarding
> >>>>>>>>>methodology and probably software tools. But I feel that
> >>>>>>>>there is more
> >>>>>>>>>to ontologies than what meets the eye of the casual
> >>>>>>user. Example:
> >>>>>>>>>I've been following developments in HL7 v3 and am every
> >>>>>>day more
> >>>>>>>>>convinced that v3 should be redesigned using readily
> >>>>>>>>available tools
> >>>>>>>>>such as Protégé (I don't think it shoud be
> >>thrown away as
> >>>>>>>>others have
> >>>>>>>>>suggestes, as the RIM contains a lot of useful domain
> >>>>>>>>information, but
> >>>>>>>>>the methodolgy user does not seem adequate).
> >>>>>>>>>
> >>>>>>>>>So, I personally feel that ontologies and Protégé-type
> >>>>>>tools could
> >>>>>>>>>turn out to be a small revolution in systems
> >>>>design. Should my
> >>>>>>>>>idea prove right, we are now in a sort of
> >>pre-paradigmatic
> >>>>>>>>situation as far
> >>>>>>>>>as application of ontologies is concerened.
> >>>>>>>>>
> >>>>>>>>>Best regards
> >>>>>>>>>
> >>>>>>>>>>-----Original Message-----
> >>>>>>>>>>From: [hidden email]
> >>>>>>>>>>[mailto:[hidden email]]
> >>>>>>On Behalf Of
> >>>>>>>>>>sridevi polavaram
> >>>>>>>>>>Sent: Monday, July 17, 2006 12:52 PM
> >>>>>>>>>>To: [hidden email]
> >>>>>>>>>>Subject: [protege-discussion] I want some solid real
> >>>>>>>>world examples!
> >>>>>>>>>>
> >>>>>>>>>>Hello All,
> >>>>>>>>>>  Please help me find an answer to my questions. I
> >>>>>>suppose some
> >>>>>>>>>>of you might also have the same questions. so let's
> >>>>>>brainstorm.
> >>>>>>>>>>I totally understand what ontology is
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>
> >>>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>>>>>--
> >>>>>>>>>--- To unsubscribe go to
> >>>>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>------------------------------------------------------------
> >>>>>>>>--
> >>>>>>>>-----------
> >>>>>>>>To unsubscribe go to
> >>>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>>>--
> >>>>>>>--- To unsubscribe go to
> >>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>--------------------------------------------------------------
> >>>>>>-----------
> >>>>>>To unsubscribe go to
> >>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>--
> >>>>>--- To unsubscribe go to
> >>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>
> >>>>>
> >>>>
> >>>>--------------------------------------------------------------
> >>>>-----------
> >>>>To unsubscribe go to
> >>>>http://protege.stanford.edu/community/subscribe.html
> >>>>
> >>>
> >>>
> >>----------------------------------------------------------------------
> >>>--- To unsubscribe go to
> >>>http://protege.stanford.edu/community/subscribe.html
> >>>
> >>>
> >>
> >>--------------------------------------------------------------
> >>-----------
> >>To unsubscribe go to
> >>http://protege.stanford.edu/community/subscribe.html
> >>
> >
> >-------------------------------------------------------------------------
> >To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
> >
> >
> >-------------------------------------------------------------------------
> >To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
> >
>
> -------------------------------------------------------------------------
> To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

Reply | Threaded
Open this post in threaded view
|

Re: Use of ontologies in software development

NEHEMIAH MAVETERA
In reply to this post by NEHEMIAH MAVETERA

Hi Nuno

Thanks. This seems like it. Will work on it. I am starting to have a direction now.

nehemiah

Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA; MSAICSIT; (SA))
Senior Lecturer- Information Systems
North West University
www.uniwest.ac.za
Tel: +27183892143
Fax: +27183892090

Can we isolate reason, and is it, so regarded, an independent source of concepts and judgements which spring from it alone? From  I. Kant, Critique of Pure Reason, A305-B362
>>> [hidden email] 07/21/06 12:09 PM >>>

Hello - Not sure if this relevant, but here goes anyway...

There is a project called ReBuilder
(http://ailab.dei.uc.pt/view_project.php?id_p=5 ) that does make use of
ontologies for software design. The project started out using WordNet (a
lexical ontology) but is now moving towards domain specific ontologies in
its second version.

--
Nuno Seco

> -----Original Message-----
> From: [hidden email] [mailto:protege-
> [hidden email]] On Behalf Of Leslie Barrett
> Sent: quinta-feira, 20 de Julho de 2006 18:30
> To: [hidden email]
> Subject: [protege-discussion] Re: Use of ontologies in software
> development
>
>
>
> Hello -
>
> Although it ihas fallen out of favor recently, WordNet was called at
> runtime
> for query expansion in various question-answering systems.
>
> Leslie Barrett
>
>
> --------------------------------------------------------------------------
> --
> From:  "NEHEMIAH MAVETERA" <[hidden email]>
> Reply-To:  [hidden email]
> To:  <[hidden email]>, <[hidden email]>
> Subject:  [protege-discussion] Use of ontologies in software development
> Date:  Thu, 20 Jul 2006 16:07:20 +0200
> >
> >Dear All
> >
> >Is there anyone who has used ontologies in software development. I mean
> at
> run time.  Please furnish me with any projects that you have done in this
> regard.
> >
> >My problems are that, as a systems developer, analyst you name it, what
> should I look for in a software development project that can be enhanced
> using ontologies.
> >
> >A typical example is " Use of ontologies in software development for
> agent
> mediated electronic marktes" . How do I make use of ontologies at run
> time?
> >
> >Nehemiah Mavetera(PhD(IT, UP, current); PM (NL); BSc (Hons) (ZW); MCSSA;
> MSAICSIT; (SA))
> >Senior Lecturer- Information Systems
> >North West University
> >www.uniwest.ac.za
> >Tel: +27183892143
> >Fax: +27183892090
> >
> >Can we isolate reason, and is it, so regarded, an independent source of
> concepts and judgements which spring from it alone? From  I. Kant,
> Critique
> of Pure Reason, A305-B362
> >>>>[hidden email] 07/20/06 2:13 AM >>>
> >
> >Well, probably someone else could answer that better than me, but I feel
> >there are some comments I would venture to make.
> >
> >An ontology for a neuron should imply an ontology for a cell. So, if
> >properly designed, should contain a potential for a complete cellular and
> >subcellular ontology. This is, to the best of my knowledge, what has been
> >done in the Formal Model of Anatomy. In the context of our project, which
> I
> >mentioned earlier, we have used Digital Anatomist for modelling anatomy
> as
> >it pertains to surgical procedures. So, although I am streching the
> concept,
> >in a certain sense an ontology for cells is part of our project. We have
> not
> >as yet started with the ontology of laboratory procedures, but when we
> reach
> >that point I am sure we will be using the cell ontology which you
> indirectly
> >mention, proably drawn form Digital Anatomist.
> >
> >Hope my comments have helped.
> >
> >>-----Original Message-----
> >>From: [hidden email]
> >>[mailto:[hidden email]] On Behalf
> >>Of sridevi polavaram
> >>Sent: Wednesday, July 19, 2006 2:29 PM
> >>To: [hidden email]
> >>Subject: [protege-discussion] Re: I want some solid real
> >>world examples!
> >>
> >>Alan,
> >>  Let's say that i have an owl description of the ontology for
> >>Neuron (cell structures in brain that processes information).
> >>Now what? how does this owl file help me? according to my
> >>understanding, i know that having an owl file is like
> >>specifying the semantics of Neuron concept.
> >>Tell me what's the first use that comes to your mind seeing
> >>an owl file. you can probably use ur own example.
> >>-Thanks
> >>Sri
> >>
> >>On 7/18/06, Alan March <[hidden email]> wrote:
> >>>
> >>>
> >>>Yep. In our case its SQL Server. What I meant by "RDF" is
> >>that we use
> >>>a table which contains the info that would otherwise be
> >>contained in
> >>>the RDF/OWL File. Sorry, I loosely called it "RDF table" but that's
> >>>just a local designation we use to refer to that table (the "rdf"
> >>>table"...granted: it does no service to proper understanding of the
> >>>concept...true reification in the literal sense of the
> >>term). You can
> >>>actually call it any name in the JDBC implementation of Protégé.
> >>>
> >>>Regarding the use of "list": An adjacency list is a manner
> >>in which to
> >>>represent a graph structure and it can be implemented in a
> >>database table.
> >>>It's a classical implmentation of graph structures. Joe Celko has
> >>>written extensively on that, using ANSI-SQL.
> >>>
> >>>Finally, there is a lot of literature on how to map (that is,
> >>>represent) RDF structures to a RDBMS. The authors of the
> >>JDBC backend
> >>>made it simple and I actually think they did it very well. Its
> >>>supposed inefficiency can be dealt
> >>>
> >>>A word of advice: if you use SQL server as a backend, don't use
> >>>Microsoft JDBC driver. I think there's an issue regarding the Java
> >>>runtime, but it simply hangs when you try to use a reasoner, with a
> >>>message regarding "column 6" which I frankly did not
> >>research into. I
> >>>just dumped the driver.
> >>>
> >>>Regards,
> >>>
> >>>Alan
> >>>
> >>>>-----Original Message-----
> >>>>From: [hidden email]
> >>>>[mailto:[hidden email]] On Behalf Of
> >>>>sridevi polavaram
> >>>>Sent: Tuesday, July 18, 2006 3:23 PM
> >>>>To: [hidden email]
> >>>>Subject: [protege-discussion] Re: I want some solid real world
> >>>>examples!
> >>>>
> >>>>Alan,
> >>>>when you say database backend, i am assuming it is some
> >>RDBMS like
> >>>>mysql, oracle etc., but you are referring it as RDF table
> >>and a list
> >>>>structure! so what kind of DB is this! can u maintain owl or RDF
> >>>>DBs!
> >>>>what is the size limit of these DBs? and how exactly are you
> >>>>achieving data storage and retrieval methods!
> >>>>-regards,
> >>>>Sri.
> >>>>
> >>>>On 7/18/06, Alan March <[hidden email]> wrote:
> >>>>>
> >>>>>
> >>>>>Well, one of the first problems we run into is OWL itself.
> >>>>Although it
> >>>>>is very powefull in expressivity, we believe it is not
> >>>>efficient as a
> >>>>>storage and retrieval repository (indeed, it was probably
> >>>>not designed
> >>>>>with efficiency in mind, and in that sense it reminds
> >>me of my old
> >>>>>Prolog days).
> >>>>>So we used the JDBC database backend. The database backend
> >>>>is not that
> >>>>>efficient in itself (and, again, its creators mention in the
> >>>>>documentation that efficiency was not one of their
> >>>>requirements) but
> >>>>>it is extremely elastic. Stated simply, the DB backend is a
> >>>>(sort of)
> >>>>>adjacency list with the particularly interesting
> >>>>characteristic that
> >>>>>the list is double linked.
> >>>>>From what I gather, one important difference with the pure RDF
> >>>>>approach is that nodes are loaded on demand, and there
> >>is a slight
> >>>>>performance issue there, but then, it allows for managing large
> >>>>>ontologies. So all design is managed using Protégé as
> >>the front-end.
> >>>>>
> >>>>>The advantage of the database backend is that it makes all the
> >>>>>elements of the ontology easily accesible for other
> >>>>applications when
> >>>>>these applications aren't involved with the design task itself.
> >>>>>Example: the multiple classifications built by reasoners such as
> >>>>>Fact++ may be accessed from web pages and applications using
> >>>>>Fact++ simple
> >>>>>approaches such as treeviews and the like, according to user's
> >>>>>different classification requirements (a CFO is interested
> >>>>if a cost
> >>>>>based classification of procedures, whilst a chief medical
> >>>>officer is
> >>>>>interested in a body systems classification of procedures).
> >>>>>Another
> >>>>>example: extracting dimensions for a warehouse. And of course
> >>>>>there are more.
> >>>>>
> >>>>>Most of these tasks are carried out by writing queries to the
> >>>>>database. We use stored procedures, which are included
> >>in the same
> >>>>>database  that contains the main "RDF" table. The pattern
> >>>>>generally used is a self linked table for accessing the
> >>adjacency
> >>>>>list schema
> >>>>>
> >>>>>Don't know how much "state of the art" this is, but it
> >>works for us.
> >>>>>
> >>>>>Regards,
> >>>>>
> >>>>>AM
> >>>>>
> >>>>>>-----Original Message-----
> >>>>>>From: [hidden email]
> >>>>>>[mailto:[hidden email]]
> >>On Behalf Of
> >>>>>>sridevi polavaram
> >>>>>>Sent: Tuesday, July 18, 2006 9:35 AM
> >>>>>>To: [hidden email]
> >>>>>>Subject: [protege-discussion] Re: I want some solid
> >>real world
> >>>>>>examples!
> >>>>>>
> >>>>>>Alan,
> >>>>>>  Thankyou very much for explaining things so
> >>patiently. So, how
> >>>>>>exactly are you using the ontology that you have tailored
> >>>>for your
> >>>>>>purpose at the application level? are you mapping the
> >>ontology
> >>>>>>properties as attributes at the database-level?
> >>>>>>if so, what are the other kinds of mapping that u do?
> >>>>>>are u using any logic based programming languages? How is
> >>>>the rich
> >>>>>>annotation that's inherent in ontologies is helping
> >>you at the
> >>>>>>application level, apart from the fact that it serves as a
> >>>>>>reinforcement knowldge base for humans. how are you
> >>>>extracting and
> >>>>>>storing information using ontologies?  can you say that
> >>>>ontology is
> >>>>>>crucial for your application and that it's not
> >>replacable by any
> >>>>>>other design methodologies.
> >>>>>>I am sorry but my mind is full of questions...just can't stop
> >>>>>>the flow..i am reading all the related literature that's
> >>>>>>available (white papers, tools, theroies, real examples,
> >>>>>>technology
> >>>>etc.,) but
> >>>>>>none of them help me converge my understanding.
> >>>>>>-regards,
> >>>>>>Sri
> >>>>>>
> >>>>>>On 7/17/06, Alan March <[hidden email]> wrote:
> >>>>>>>
> >>>>>>>
> >>>>>>>>>Wow SNOMED and UMLS are supposingly success stories for
> >>>>>>>>>ontology
> >>>>>>>>applications. and they failed to mee your requirements!!
> >>>>>>>>then the whole point of "sharing knowldge" seems
> >>skeptical.
> >>>>>>>>So are u designing ontologies from scratch for your HMO
> >>>>>>application!
> >>>>>>>>everyone is apparently having there own ontology
> >>>>>>>
> >>>>>>>The trouble we faced with SNOMED and UMLS is they
> >>>>appeared to us
> >>>>>>>as incomplete as far as attributes (relations) are
> >>>>concerned: too
> >>>>>>>few, failing to capture the richness of possible relations.
> >>>>>>>And then of course, the mistakes found in them, which Barry
> >>>>Smith and
> >>>>>>others have
> >>>>>>>mentioned in their papers. Also, much of the terms we had
> >>>>>>to use were
> >>>>>>>highly local and neither SNOMED nor UMLS offered a clear
> >>>>>>>method for extensions. Although SNOMED has
> >>something on those
> >>>>>>>lines,
> >>>>>>the problem
> >>>>>>>of its "ontological incompleteness"
> >>>>>>>and licensing questions represented a drawback for us. But
> >>>>>>then again:
> >>>>>>>both
> >>>>>>>SNOMED and UMLS have been extremely usefull to us
> >>as a sort of
> >>>>>>>knowledge repositories, but we just could not use them
> >>>>>>"out-of-the-box".
> >>>>>>>
> >>>>>>>Actually, we didn't start from scratch: we reutilized
> >>>>much of the
> >>>>>>>publically available Galen files, and modelled our ontology
> >>>>>>following
> >>>>>>>an upper level ontology. For anatomy we used Digital
> >>>>Anatomist as
> >>>>>>>published in UMLS. We tried to use as much as possible from
> >>>>>>>several ontologies we found to be suitable to our purposes.
> >>>>>>>But
> >>>>>>yes, we had to
> >>>>>>>reinvent the wheel in some circumstances. In any case, I am
> >>>>>>not quite
> >>>>>>>sure if we will ever manage to arrive at a universally
> >>>>>>>accepted ontology. Probably at the upper and middle levels
> >>>>>>>this will be achieved, but I doubt it will work for "bottom"
> >>>>>>>levels.
> >>>>>>>
> >>>>>>>Regarding success stories: well, I guess SNOMED is a
> >>>>>>success story in
> >>>>>>>a certain sense, but I am not quite sure if it can be
> >>>>>>called that way
> >>>>>>>as far as ontological principles are concerned. Not to deny
> >>>>>>its value:
> >>>>>>>it is rich in content. Worth mentioning is the fact
> >>>>that we didn't
> >>>>>>>want to consider terminologies which weren't in the public
> >>>>>>domain or commercial in nature.
> >>>>>>>I
> >>>>>>>personally feel that medical terminologies such be
> >>built in a
> >>>>>>>collaborative manner, probable on the same lines as Linux
> >>>>>>was made (by
> >>>>>>>having different groups managing different parts of the
> >>>>domain).
> >>>>>>>In that sense, I feel that Rector's advice regarding the way
> >>>>>>ontologies
> >>>>>>>should be built (by
> >>>>>>>"untangling") allows for this. I am relly sorry that the
> >>>>>>Galen project
> >>>>>>>didn't continue. Looking back on it, I think it is
> >>>>still the best
> >>>>>>>thing we have seen in all these decades of discussion on
> >>>>>>terminologies
> >>>>>>>and how they should be built and maintained.
> >>>>>>>
> >>>>>>>>
> >>>>>>>>In that sense, I find that ontologies are "the" techique
> >>>>>>for design,
> >>>>>>>>>development and maintaince of so-called controlled
> >>>>>>terminologies.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>>yes, i would say, that even i am convinced that at
> >>>>the level
> >>>>>>>>>of
> >>>>>>>>vocabularies. like a dictionary sort of thing. the
> >>>>wikipedia for
> >>>>>>>>encyclopedia also uses ontologies!!
> >>>>>>>>Thanks a lot for giving me some idea about the real workd
> >>>>>>scenario.
> >>>>>>>>Best,
> >>>>>>>>-Sri
> >>>>>>>>
> >>>>>>>>OF course,
> >>>>>>>>>in the process of using ontologies one quickly realizes
> >>>>>>>>>the
> >>>>>>>>fact that
> >>>>>>>>>they could probably be usefull for analysis and design
> >>>>>>of complex
> >>>>>>>>>systems, and I personally feel that they could
> >>>>>>ultimately replace
> >>>>>>>>>other tools such as UML and the like. This, of
> >>course (and
> >>>>>>>>at least as
> >>>>>>>>>far as my comprehension goes), still requires much work
> >>>>>>regarding
> >>>>>>>>>methodology and probably software tools. But I feel that
> >>>>>>>>there is more
> >>>>>>>>>to ontologies than what meets the eye of the casual
> >>>>>>user. Example:
> >>>>>>>>>I've been following developments in HL7 v3 and am every
> >>>>>>day more
> >>>>>>>>>convinced that v3 should be redesigned using readily
> >>>>>>>>available tools
> >>>>>>>>>such as Protégé (I don't think it shoud be
> >>thrown away as
> >>>>>>>>others have
> >>>>>>>>>suggestes, as the RIM contains a lot of useful domain
> >>>>>>>>information, but
> >>>>>>>>>the methodolgy user does not seem adequate).
> >>>>>>>>>
> >>>>>>>>>So, I personally feel that ontologies and Protégé-type
> >>>>>>tools could
> >>>>>>>>>turn out to be a small revolution in systems
> >>>>design. Should my
> >>>>>>>>>idea prove right, we are now in a sort of
> >>pre-paradigmatic
> >>>>>>>>situation as far
> >>>>>>>>>as application of ontologies is concerened.
> >>>>>>>>>
> >>>>>>>>>Best regards
> >>>>>>>>>
> >>>>>>>>>>-----Original Message-----
> >>>>>>>>>>From: [hidden email]
> >>>>>>>>>>[mailto:[hidden email]]
> >>>>>>On Behalf Of
> >>>>>>>>>>sridevi polavaram
> >>>>>>>>>>Sent: Monday, July 17, 2006 12:52 PM
> >>>>>>>>>>To: [hidden email]
> >>>>>>>>>>Subject: [protege-discussion] I want some solid real
> >>>>>>>>world examples!
> >>>>>>>>>>
> >>>>>>>>>>Hello All,
> >>>>>>>>>>  Please help me find an answer to my questions. I
> >>>>>>suppose some
> >>>>>>>>>>of you might also have the same questions. so let's
> >>>>>>brainstorm.
> >>>>>>>>>>I totally understand what ontology is
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>
> >>>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>>>>>--
> >>>>>>>>>--- To unsubscribe go to
> >>>>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>------------------------------------------------------------
> >>>>>>>>--
> >>>>>>>>-----------
> >>>>>>>>To unsubscribe go to
> >>>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>>>--
> >>>>>>>--- To unsubscribe go to
> >>>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>--------------------------------------------------------------
> >>>>>>-----------
> >>>>>>To unsubscribe go to
> >>>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>
> >>--------------------------------------------------------------------
> >>>>--
> >>>>>--- To unsubscribe go to
> >>>>>http://protege.stanford.edu/community/subscribe.html
> >>>>>
> >>>>>
> >>>>
> >>>>--------------------------------------------------------------
> >>>>-----------
> >>>>To unsubscribe go to
> >>>>http://protege.stanford.edu/community/subscribe.html
> >>>>
> >>>
> >>>
> >>----------------------------------------------------------------------
> >>>--- To unsubscribe go to
> >>>http://protege.stanford.edu/community/subscribe.html
> >>>
> >>>
> >>
> >>--------------------------------------------------------------
> >>-----------
> >>To unsubscribe go to
> >>http://protege.stanford.edu/community/subscribe.html
> >>
> >
> >-------------------------------------------------------------------------
> >To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
> >
> >
> >-------------------------------------------------------------------------
> >To unsubscribe go to http://protege.stanford.edu/community/subscribe.html
> >
>
> -------------------------------------------------------------------------
> To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html


-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html

Reply | Threaded
Open this post in threaded view
|

Re: Use of ontologies in software development

Jennifer Vendetti-2
In reply to this post by NEHEMIAH MAVETERA

Nehemiah,

NEHEMIAH MAVETERA wrote:

>Is there anyone who has used ontologies in software development. I mean at run time.  Please furnish me with any projects that you have done in this regard.
>  
>
We have a "Projects that use Protege" section on our Wiki that you may
find useful:

http://protege.cim3.net/cgi-bin/wiki.pl?ProjectsThatUseProtege

>My problems are that, as a systems developer, analyst you name it, what should I look for in a software development project that can be enhanced using ontologies.
>
>A typical example is " Use of ontologies in software development for agent mediated electronic marktes" . How do I make use of ontologies at run time?
>  
>
Protege has a Java API for accessing ontologies built with Protege.  It
is documented on our Web site:

http://protege.stanford.edu/doc/dev.html#api

Jennifer


-------------------------------------------------------------------------
To unsubscribe go to http://protege.stanford.edu/community/subscribe.html