Re: linking a symbol with a dataset

Hi Vladimir,
 As you say, nothing is atomic. Even the symbols in the s,p,o slots
represent entities that may very well be divisible. In SIO, we
developed a property called 'has direct part' [1], which enables
modelers to specifically select objects of interest. While not exactly
what you suggested, is this closer to what you think is needed?

m.

[1] http://semanticscience.org/resource/has-direct-part
Michel Dumontier
Associate Professor of Medicine (Biomedical Informatics), Stanford University
Chair, W3C Semantic Web for Health Care and the Life Sciences Interest Group
http://dumontierlab.com


On Tue, Dec 9, 2014 at 2:16 PM, Vladimir Mironov
<vladimir.n.mironov@gmail.com> wrote:
> Hi Michel,
>
> as everybody knows, atomic means 'indivisible'. In the world of Nature, of
> course, there is nothing atomic.
> However, in the world of RDF subject, predicate, and object are atomic -
> there are no other concepts these three could be subdivided into.
> I haven't seen such a property so far in vocabularies, I proposed it because
> you guys could not find anything suitable for the task.
> I feel it serves the purpose and nothing may prevent you from introducing a
> new property in want of appropriate ones. This one should be, of course, a
> subProperty of 'partOf'.
>
> Cheers
>
> On 9 December 2014 at 19:36, Michel Dumontier <michel.dumontier@gmail.com>
> wrote:
>>
>> Hi Vladimir,
>>   Can  you elaborate with a definition for 'isAtomicPartOf'?  Is this
>> already defined in a vocabulary?
>>
>> m.
>> Michel Dumontier
>> Associate Professor of Medicine (Biomedical Informatics), Stanford
>> University
>> Chair, W3C Semantic Web for Health Care and the Life Sciences Interest
>> Group
>> http://dumontierlab.com
>>
>>
>> On Tue, Dec 9, 2014 at 12:33 AM, Vladimir Mironov
>> <vladimir.n.mironov@gmail.com> wrote:
>> > Hi,
>> >
>> > how about 'isAtomicPartOf'?
>> >
>> > Vladimir
>> >
>> > On 8 December 2014 at 18:04, Michel Dumontier
>> > <michel.dumontier@gmail.com>
>> > wrote:
>> >>
>> >> Hi all,
>> >>   On the call today we discussed the issue [1] of linking a symbol in
>> >> a triple (e.g. a subject, a predicate, *or* an object) to a dataset.
>> >> The use case for this is twofold : to provide a direct link between
>> >> data items and their datasets in a Linked Data manner, and to survey
>> >> the use of data items across datasets. While we agreed that using a
>> >> relation such as dc:isPartOf is fairly natural to link the triple
>> >> itself to the dataset, it is much less clear for linking the
>> >> components to the dataset. In Bio2RDF we used void:inDataset, but the
>> >> domain of this relation is a foaf:Document, so it muddies the
>> >> semantics by entailing a possible disjoint type with whatever the
>> >> subject has been typed with (e.g. protein, disease, etc).
>> >>
>> >> We discussed the suitability of existing vocabularies, but none, to
>> >> our knowledge, clearly fit the situation. For instance, can
>> >> dc:isPartOf (http://purl.org/dc/terms/isPartOf) be used as a logical
>> >> partition of the dataset with the data item? or is SIO's refers to
>> >> (http://semanticscience.org/resource/refers-to) potential suitable, if
>> >> not somewhat vague?
>> >>
>> >> We welcome your thoughts on the matter. Do you know of a suitable
>> >> relation? Should we consider some new relation such as utilizes /
>> >> is-utilized-in or is-data-item-in / has-data-item?
>> >>
>> >> Cheers!
>> >>
>> >> m.
>> >>
>> >> [1] https://github.com/joejimbo/HCLSDatasetDescriptions/issues/90
>> >>
>> >> Michel Dumontier
>> >> Associate Professor of Medicine (Biomedical Informatics), Stanford
>> >> University
>> >> Chair, W3C Semantic Web for Health Care and the Life Sciences Interest
>> >> Group
>> >> http://dumontierlab.com
>> >>
>> >
>
>

Received on Thursday, 11 December 2014 19:20:39 UTC