ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Semantic Enterprise Architecture - Interoperability?

To: ontolog-forum@xxxxxxxxxxxxxxxx
From: "doug foxvog" <doug@xxxxxxxxxx>
Date: Wed, 8 Sep 2010 17:12:03 -0400 (EDT)
Message-id: <55365.71.163.21.40.1283980323.squirrel@xxxxxxxxxxxxxx>
On Wed, September 8, 2010 13:04, Pavithra said:
> Doug,
> Thanks for the detail explanation of how such concepts are resolved in
> Cyc.  I agree when Ontology is developed using controled and well defined
> terms,  it should help.
>
> But Web is full of user developed/ filled content.      (01)

Which is why a Semantic Web needs ontologies of terms with fixed meanings.    (02)

> My issue was more like term "Gas" as in gasoline and petrol  vs Gas as in
> Oxygen and Nitrogen  in the same country, along with different countries.
> One is a derived usage where as other is the actual term.    (03)

These seem to me to be NLP issues, not ontological issues.  Be that as it
may, I'd say that both "Gas"es are actual terms.  Similarly, "oxygen" and
"nitrogen" are words for both elements and binary compounds which are
gasses at standard temperature and pressure.    (04)

> Your example AmericanEnglishNT  would have these two words with total
> different meaning.  Because here we do use the word Gas as in Oxygen and
> Nitrogen.    If you do not know the meaning of the word,  how would you
> solve it?    (05)

I assume you mean that i don't know which of a set of meanings to select
when doing NLP.  I would solve the problem as follows:
* Create several contexts for different sets of selected meanings of terms.
* If the dialect (British, American, Australian, ...) is known, restrict
  possible meanings to those in that dialect's context.
* Generate assertions matching the NL statements using predicates with
  type-restricted arguments.
* Reject interpretations that would make the generated assertion invalid.
  E.g.
  Oxygen is a gas.  The party was a gas.  I filled my Prius with gas.
  The first denotation for Gas-TheWord would be used for sentence 1.
  The fifth denotation for Gas-TheWord would be used for sentence 2.
  The second denotation for Gas-TheWord(in Am.EnglishMt) would be for 3.    (06)

> Along with URI reference, there is URN  (Unified Reference Name) which
> should give more of a granular reference.   However is that effective? 
> Does  it solve the above example of multiple meanings?     (07)

So long as the different otological terms have different URIs, there are
no multiple meanings of the ontological terms.  It is an NLP issue to map
the English terms (with multiple meanings) to the correct ontological term.    (08)

> When I worked for Citigroup,  we had to defined country context - for
> money, language, conversion factors etc for systems..   I am not saying it
> is new, but it is interesting in terms of Semantic Web and Semantic
> Interoperability and all the new technologies.    (09)

The country context was useful because the data were using did not
specify their units used.  The contexts gave you that information.    (010)

-- doug    (011)


> Regards,
> Pavithra
>
> --- On Tue, 9/7/10, doug foxvog <doug@xxxxxxxxxx> wrote:
>
> From: doug foxvog <doug@xxxxxxxxxx>
> Subject: Re: [ontolog-forum] Semantic Enterprise Architecture -
> Interoperability?
> To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
> Date: Tuesday, September 7, 2010, 5:10 PM
>
> One problem that often occurs in discussing computer ontologies is the
> conflation of words, terms in an ontology, and the things represented
> by those words and terms.  If one stops to think, it is clear that these
> are different things.
>
> Words have conjugations, declinations, derivations, plurals,
> pronunciations, spellings, and meanings.  Terms in an ontology do not
> have linguistic properties.  The things represented by the words and
> ontological terms may have have physical properties, which the other
> two types of object don't have.
>
> To discuss meanings of words and terms in ontologies, it is necessary
> to carefully distinguish these three classes of things.
>
> On Mon, September 6, 2010 10:29, Pavithra said:
>> ...
>> Semantic interoperability is new and mature ontology development model
>> would help create that.  But there are always words that are used with
>> different semantics based on context, culture, environment.    
>
> Semantic interoperability should not be based on words, but on defined
> terms with fixed definitions.  In computer languages it isn't an issue
> that reserved words (object, method, ...) have different meanings in
> other contexts.  Terms in ontologies should be similarly be treated
> as reserved words and thus be unambiguous.
>
> The combination of a specified ontology with a specified term in the
> ontology has a unique meaning.  A different ontology which has a term
> with the same name should cause no problems so long as the ontology
> which is used when the term is referenced is clear.
>
> URIs provide such unambiguity by having a leading part which indicates
> the ontology & a final part which indicates the name of a term within
> that ontology.  Languages which use name spaces have a method of
> indicating the name space (representing the ontology) connected to the
> name within that name space.  These can be translated into URIs by
> appending the term name to the URI representing the name space.
>
>> In America
>> a gas tank holds gas which is liquid..  ( yeah gas derived from the word
>> gasoline for the word petrol. ) 
>
> It is the *word* "gasoline" which is used in the US for the substance
> denoted (in the UK & Ireland) by the *word* "petrol".  Of course the
> word "petrol" is derived from the word "petroleum" -- but in a different
> fashion that the physical petrol is derived from physical petroleum. 8)#
>
>> But if you use the web in another
>> country  one is looking at  gas, which has totally different properties
>> then liquid.  I mean where is it defined that gas = petrol?  
>
> If you are referring to the meaning of the words, the question could be
> more clearly phrased: "where is it defined that the meanings of the
> words spelled "gas" and "petrol" are the same?
>
> In Cyc, six statements map the words spelled "gas", "gasoline", and
> "petrol" to the concept of gasoline, three mapping the text strings
> to conceptual entities called words, and three providing denotations of
> the two words which are the same.
>
> in the context EnglishMt:
> (denotation Gasoline-TheWord 1 MassNoun GasolineFuel).
> (massNumber Gasoline-TheWord "gasoline").
>
> in the context AmericanEnglishMt:
> (denotation Gas-TheWord 2 MassNoun GasolineFuel).
> (massNumber Gas-TheWord "gas").
>
> in the context BritishEnglishMt:
> (denotation Petrol-TheWord 1 MassNoun GasolineFuel).
> (massNumber Petrol-TheWord "petrol").
>
> In the context BritishEnglishMt (or contexts which inherit it) Cyc
> can conclude that the words spelled "petrol" and "gasoline" can each
> refer to the same thing.  This is because BritishEnglishMt inherits
> EnglishMt.  Likewiae, in AmericanEngishMt (which also inherits EnglishMt)
> Cyc can conclude that the words spelled "gas" and "gasoline" can each
> refer to the same thing.
>
> However, to conclude that "gas" and "petrol" refer to the same thing,
> Cyc would have to be in a context which inherits both AmericanEnglishMt
> and BritishEnglishMt.  Such a context should not be used in language
> processing.
>
>> Then there
>> are same words that are used in different cultures and different
>> languages..  
>
> Similarly, the mapping of words and meanings should be defined in
> contexts particular to the appropriate culture and/or language.
>
>> There are some proposed solutions -  an Ontology repository that tracks
>> semantics to use it to harmonize meanings could help to resolve such
>> semantic difference.     
>
> It appears that you are referring to using ontologies for natural language
> processing and understanding here.
>
>> The same solutions would work for Semantic web too..
>
> If full URIs were used on the Semantic Web, it would not be an issue
> since the different meanings of terms would use different URIs.
>
>>  However at present the concept of Linked data seems more at URI and
>> URL level.. kind of at object level rather than words and things. 
>
> As discussed at the beginning, words and things are at two very different
> levels.  And computer ontology terms for describing them is a third
> distinct level.
>
> -- doug f
>
>> ( I am
>> not sure , I am trying say the granularity could be different). which is
>> again structural level. 
>>
>> Thanks,
>> Pavithra
>
> ============================================================doug foxvog   
> doug@xxxxxxxxxx   http://ProgressiveAustin.org
>
> "I speak as an American to the leaders of my own nation. The great
> initiative in this war is ours. The initiative to stop it must be ours."
>     - Dr. Martin Luther King Jr.
> ===========================================================
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/ 
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/ 
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
> To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
>
>
>
>
>    (012)


=============================================================
doug foxvog    doug@xxxxxxxxxx   http://ProgressiveAustin.org    (013)

"I speak as an American to the leaders of my own nation. The great
initiative in this war is ours. The initiative to stop it must be ours."
    - Dr. Martin Luther King Jr.
=============================================================    (014)


_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (015)

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