W3C home > Mailing lists > Public > semantic-web@w3.org > June 2007

Re: part-whole semantic relationship

From: David Price <david.price@eurostep.com>
Date: Tue, 12 Jun 2007 17:04:31 +0100
To: semantic-web@w3.org
Message-Id: <200706121704.31943.david.price@eurostep.com>

I missed the early discussions on this but there seem to be some basic 
misunderstandings here - or at least wrt the engineering disciplines.

In the engineering world, "whole-part" and "whole surrounding part" are not 
the same concept. The wing of a plane is part of the plane but is not 
surrounded by the plane. "located within" and "part of" are two different 
relationships. Ships and Oil Platforms are often broken down by Zones to 
locate items, but that's not the same part-of relationship used to build 
them.

The example of John being a Football team also is a misunderstanding of the 
part-of semantics. John is a member-of a Football team, not a part-of the 
team. It's a different relationship because part-of is a transitive 
relationship. Any part-of John (i.e. his hand) is not a part of the team and 
as part-of is transitive, John is clearly not part-of the team either.

"Assembly relationships", which is what this seems to really be about are well 
understood in the engineering world. The comment that these are all just 
Parts that have other Parts is how it's handled there. Identifying what 
precisely is a "Part" is left to the individual organization though because 
the builder of a component assembly sees their product as an assembly while 
the users of that assembly in a larger system just see a Part. Nobody's 
wrong, it's just a matter of perspective. In any case, the systems that 
manage this stuff in engineering companies have been around 20-30 years or 
more. They're now called PDM (Product Data Management) or PLM (Product 
Lifecycle Management) systems. Have a look at one.

Cheers,
David

On Tuesday 12 June 2007 16:18:17 Rikkert Koppes wrote:
> There is a lot of redundancy here. Nesting depth should be infered
> rather than specified if you ask me. superPart and subPart are already
> just parts. Whether they are sub- or super parts can be inferred from
> the containment structure. Suppose you have a superpart defined, a
> camera say and you wrap it in a box, which you put in a sea container.
> This would make the camera a sub part again. Therefore, part as a type
> alone would just do.
>
> How do sequence id's differ from rdf:seq types?
>
> I think most of your ??id's really prevent expansion. If you want to do
> this, use rdf containres instead.
>
> Rikkert Koppes (mophor)
>
> tianxama wrote:
> >     Part-Whole relationships. Part(s) describing surronding Whole.
> >
> > The idea is to develop kind of data structure, with parts nested
> > compounding a whole, and the inner parts of the structure knows, in
> > reverse order, the containing tree of elements or structure around it.
> >
> >
> >       1: Definitions:
> >
> >    1. *superPart:* Part containing one or more parts.
> >    2. *part:* Inside a superPart. Can contain subParts.
> >    3. *subPart:* Inner unit inside a part.
> >    4. *NID:* Nesting level ID (deep level entities are in a
> >       nested/tree structure)
> >
> >       XML Example:
> >
> >          <superPart NID="0">
> >            <part NID="1">
> >              <subPart NID="2">
> >            </part> <!-- end NID 1 -->
> >          </superPart> <!-- end NID 0 -->
> >
> >    5. *SeqID:* Sequence ID (symbol sequencing and identification)
> >
> >       Symbols sequencing and identification begins with first
> >       occurrence of a symbol in a structure, begining with 0, next
> >       symbol is tagged with 1. If next symbol appeared before, then
> >       the tag used before to identify it is used.
> >
> >    6. *OccID:* Occurrence ID
> >
> >       Times element appeared in a sequence. First time same element
> >       occurs is 0, then 1, etc.
> >
> >    7. *CID:* Count ID (0, 1, 2, ...)
> >
> >       Sequential count, begining with 0, of elements in determinate
> >       context (Nesting level, Sequence, etc.)
> >
> >    8. *MvtID:* Movement ID
> >
> >       Signals start (S), inside (I), outside (O) or end (E) movement
> >       in a statement chain.
> >
> >
> >       2. Example:
> >
> >     * Linear Form:
> >
> >         ( superPart ( part ( ( superPart ) ( subPart ( part ( superPart )
> > ) ) ) )
> >
> > NID:         0         1           2            2        3        4
> >
> > SeqID:       0         1           0            2        1        0
> >
> > OccID:       0         0           1            0        1        2
> >
> > CID:         0         1           2            3        4        5
> >
> > MvtID:       S         I           O            I        O        E
> >
> >
> >
> >     * XML Example:
> >
> > <superPart NID="0">
> >   <part NID="1">
> >     <superPart NID="2"/>
> >     <subPart NID="2">
> >       <part NID="3">
> >         </superPart NID="4">
> >       </part>
> >     </subPart>
> >   </part>
> > </superPart>
> >
> >
> >
> >     * Another Example:
> >
> >         ( head ( face ( ( head ) ( eyes ( face ( head ) ) ) ) )
> >
> > NID:        0      1        2        2      3      4
> >
> > SeqID:      0      1        0        2      1      0
> >
> > OccID:      0      0        1        0      1      2
> >
> > CID:        0      1        2        3      4      5
> >
> > MvtID:      S      I        O        I      O      E
> >
> >
> >
> >
> >       3. Composite IDs:
> >
> > Combining CID, NID, SeqID and OccID, other IDs for easier retrieval of
> > elements in statement can be achieved.
> >
> > For example: NID + OccID = CID (Count of elements inside a specific
> > nesting level)
> >
> >         ( head ( face ( ( head ) ( eyes ( face ( head ) ) ) ) )
> >
> > NID:        0      1        2        2      3      4
> >
> > SeqID:      0      1        0        2      1      0
> >
> > OccID:      0      0        1        0      1      2
> >
> > CID:        0      1        2        3      4      5
> >
> > OrderID:  (0.0)  (1.0)    (2.0)    (2.1)  (3.0)  (4.0)
> >
> >
> >
> >
> >       4. Terms (Parts) Relationships and Roles:
> >
> > In specific context of container part, another part plays a role or
> > function in the context of container part. This is the semantics of
> > the containment relationship, for example: John, inside a specific
> > Team, is a player, more specificaly, a goalkeeper.
> >
> > Parsing Example:
> >
> > [previusElement]:[role/type] => [actualElement]:[role/type] =>
> > [nextElement]:[role/type]
> >
> >
> >
> >       5. Measurement units:
> >
> > superUnit (eg.: Kilometer)
> > unit (eg.: Meter)
> > subUnit (eg.: Centimeter)
> >
> > Ratio:
> >
> > superUnit (1) => unit (1/n) => subUnit (1/n : n)
> >
> >
> > _sebastián samaruga._
> >
> > ------------------------------------------------------------------------
> > View this message in context: part-whole semantic relationship
> > <http://www.nabble.com/part-whole-semantic-relationship-tf3896550.html#a1
> >1046368> Sent from the w3.org - semantic-web mailing list archive
> > <http://www.nabble.com/w3.org---semantic-web-f11665.html> at Nabble.com.



-- 
UK +44 20 7221 7307
Mobile +44 7788 561308
Skype +1 336 283 0606

Eurostep Limited. Registered in England and Wales No.03049099
Registered Office: Cwttir Lane, St. Asaph, Denbighshire LL17 0LQ.
Received on Tuesday, 12 June 2007 16:06:48 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:15 GMT