Let us consider Semantic Web and Linked Data as a conceptual references without 
going into full details about their meaning. Within such limited scope we can 
say that Semantic Web is about publishing knowledge on the Web in a machine 
processable way while Linked Data is about publishing data on the Web such that 
it would not be in collision with Semantic Web. Consequently, the concept of 
resource is of the paramount importance being anything identified by URI. In 
the case of Semantic Web, URI represents UID of an abstract concept while in 
the case of Linked Data it is dereferenced by means of HTTP protocol to the 
corresponding data item.

Now, having published something (concept/individual/data) by assigning an URI 
to it means nothing on its own. For example, we call our pet hamster Peach and 
it is definitely  in collision with the common meaning of the word (hamster is 
not fruit). Since, if I publish something by an URI to it means nothing until 
somebody else starts using it. This is in compliance with the open world 
assumption. So, while there is no (formal) barrier to assign any valid URI to 
any resource it would have no meaning unless other agents may access it in a 
systematically manageable way.

Analogous to writing, there is no constrain that would stop me writing a novel 
such that every single sentence is written in a separate (uniquely identified) 
textual file. However, it would be very hard and almost impossible for other 
people to read it. So, writers usually write novels such that the whole text is 
contained within a single file. So, creating individuals, that is very close 
cognitive process to writing sentences of text, should be done in a similar 
way: Include/reference to all ontologies that you need for domain context 
representation but create all individuals as well as statements within a single 
domain representing a manageable dataspace. Regardless of the potentially 
diverse domains of included ontologies, all individuals and statements about 
them stay within a single dataspace. In this way, materialized dereferencing as 
is the case in Linked Data is easily supported as well.

Milorad





>________________________________
> From: Marco Neumann <[email protected]>
>To: [email protected]; Milorad Tosic <[email protected]> 
>Sent: Thursday, August 8, 2013 9:42 AM
>Subject: Re: best practice in assigning URIs to individuals
> 
>
>what did you learn? how do you go about the assignments of URIs to
>individuals now?
>
>
>On Thu, Aug 8, 2013 at 3:35 AM, Milorad Tosic <[email protected]> wrote:
>
>> Yes, that should be sufficient.
>>
>> Thanks,
>> Milorad
>>
>>
>>
>>
>> >________________________________
>> > From: Andy Seaborne <[email protected]>
>> >To: [email protected]
>> >Sent: Wednesday, August 7, 2013 10:00 PM
>> >Subject: Re: best practice in assigning URIs to individuals
>> >
>> >
>> >On 07/08/13 10:50, Milorad Tosic wrote:
>> >> Thanks all for helping me out! I totally agree with you.
>> >>
>> >> But, is there a strong argument (backed by a standard, for example, or
>> a recommendation from a standard body ...) supporting this opinion that can
>> be used in argumentation?
>> >
>> >Maybe this will help:
>> >
>> >http://www.w3.org/TR/webarch/#uri-assignment
>> >
>> >>
>> >> Milorad
>> >
>> >    Andy
>> >
>> >
>> >
>> >
>>
>
>
>
>-- 
>
>
>---
>Marco Neumann
>KONA
>
>
>

Reply via email to