[ontoiop-forum] ontoiop_20140924: Chat Transcript - sent by: TillMossakowski

[Till Mossakowski]

*Chat transcript from room: ontoiop_20140924*
*2014-09-24 GMT-08:00*
*[07:43] **TerryLongstreth: *In a hotel room, working on getting an
audio connection...
*[08:05] *anonymous morphed into ChristophLange
*[08:06] **ChristophLange: *Hi, I'm back to OntoIOp, but unfortunately
didn't have time to prepare anything, nor to catch up with previous
things. If there is anything related to syntax or IRIs where my advice
would have been helpful, I'm happy to help though
*[08:14] **TillMossakowski: *Till presenting slightly revised slides
*[08:14] **TillMossakowski: *slide 12
*[08:17] **TillMossakowski: *physical distribution of OMS is entirely
orthogonal to the distinction made here (focused OMS, distributed
OMS=OMS networks, OMS libraries). Actually, even focused OMS can be
physically distribtued, while distributed OMS can be physically
non-distributed.
*[08:18] **ChristophLange: *@TaraAthan: BTW in parallel to this call I'm
reviewing the minutes of the last one (when I was on holiday). I'll be
happy to resolve any LoLa-related issues via the mailing list or
personal email.
*[08:19] **TillMossakowski: *The top-level definitions in OMS libraries
are similar to titlings in Common Logic documents.
*[08:19] **ChristophLange: *I see that the terminology has changed, so
I/we may need to re-apply some of these changes in LoLa
*[08:19] **ChristophLange: *We can do so via
https://github.com/ontohub/OOR_Ontohub_API/issues. End of "LoLa
diversion" :-)
*[08:20] **TillMossakowski: *Common Logic documents can mix titlings
with other assertions.
*[08:22] **TillMossakowski: *in many logics (indeed, nearly all logics
except Common Logic), titling maps and models are kept seperate.
Therefore, we do the same in DOL.
*[08:25] **TillMossakowski: *Tara: DOL should be able to support Common
Logic without nesting of titlings.
*[08:27] **TillMossakowski: *Since we do not have our hand at the models
(they could be models in any logic), we cannot assume that models
include titling maps.
*[08:31] **TillMossakowski: *Tara: it seems that only Common Logic
without titling and imports would be conformant to DOL. The DOL library
mechanism needs to be used for titling - otherwise, DOL and CL are just
incompatible.
*[08:35] **TerryLongstreth: *The foregoing discussion presupposes that
Common Logic V2 will become the current draft version, and that
implicitly, that version of CL will be incompatible with the current
draft of OntoIop
*[08:36] **TerryLongstreth: *So, do we need a harmonization session?
*[08:38] **TillMossakowski: *Tara: it could be useful that libraries
have a (hierarchical) structure
*[08:39] **TillMossakowski: *However, libraries define OMS, and thereby
assign them URIs. So you can reference the OMS by their URIs, and do not
need to include the library.
*[08:41] **TillMossakowski: *Tara: hierarchical libraries could be
useful for nested scoping and for searching.
*[08:42] **TillMossakowski: *A useful addition to DOL could be an
include statement for libraries.
*[08:43] **TillMossakowski: *Tara: there could be the need for
annotating the include statement (e.g. with date, or other metadata)
*[08:45] **TillMossakowski: *Sn include statement would import the whole
library. Note that still, the individual OMS would be referenced by
their URLs.
*[08:45] **TillMossakowski: *Also, prefix maps would be imported.
*[08:46] **TillMossakowski: *In a library, the individual OMS should
have URIs that are derived from the URI of the library and a local name
(relative URI) of the OMS.
*[08:48] **TillMossakowski: *At least, OntoIOp adopts this principle.
*[08:49] **TillMossakowski: *DOL should not adopt this principle (there
may be libraries of OMS with very different URIs).
*[08:50] **TillMossakowski: *But DOL should make it possible to adopt
this principle, via relative IRIs, or CURIEs.
*[08:51] **TillMossakowski: *If you define the emtpy prefix to be the
IRI of the library, then local names would just be appended to that IRI,
by the CURIE mechanism.
*[08:52] **TillMossakowski: *This could be a default behaviour.
*[08:54] **TillMossakowski: *In RDF turtle syntax, such CURIEs start
with a colon, in OWL Manchester syntax, they don't (and that is also the
choice that we have made in DOL).
*[08:59] **TillMossakowski: *A problem is that symbols in an OMS always
need to be referenced with oms_name/sym_name, because their IRI also
includes the OMS name.
*[09:00] **TerryLongstreth: *I have to leave. Got another meeting. Will
try to spend some time with Ch. 9.
*[09:00] **TillMossakowski: *If we do not want this, we would have to
adopt MMT IRIs, which allow for referencing symbol in their ontologies,
and ontologies in their libraries.
*[09:01] **ChristophLange: *CURIEs (part of RDFa 1.1):
http://www.w3.org/TR/rdfa-syntax/#s_curies
*[09:01] **ChristophLange: *OWL Manchester example:
http://www.w3.org/TR/owl2-manchester-syntax/#Quick_Reference
*[09:02] **TaraAthan: *We should carefully distinguish between empty
prefix (e.g. :abc) and no prefix (e.g. abc)
*[09:03] **TillMossakowski: *MMT:
https://svn.kwarc.info/repos/MMT/doc/html/index.html paper about MMT:
http://www.sciencedirect.com/science/article/pii/S0890540113000631 or
(open-access but a bit outdated) http://arxiv.org/abs/1105.0548
*[09:03] **ChristophLange: *CURIE calls :sym "default prefix", and sym
"no prefix"
*[09:09] **TaraAthan: *http://www.w3.org/TR/rdfa-syntax/ says that it is
possible to define a default prefix (through @vocab) to be applied to
"undefined names", which I believe are no prefix names.
*[09:15] **TillMossakowski: *We suggest to let the no-prefix default to
a context-sensitive expansion mechanism, which always prepends the
library IRI (in the context of a structured OMS where named OMS a
referenced) resp. the current OMS IRI (in the context of a basic OMS).
*[09:18] **TillMossakowski: *what would be the prepended IRIs exactly?
*[09:19] **TillMossakowski: *in MMT, they use
library_name?OMS_name?symbol_name
*[09:19] **TillMossakowski: *we cannot use library_name#OMS_name#symbol_name
*[09:20] **TillMossakowski: *because only one # can be used for the
fragment (in this case, the fragment would be OMS_name#symbol_name)
*[09:21] **TillMossakowski: *the main problem with the fragment (#) is
that one needs to download the whole library, which can be inefficient
*[09:22] **TillMossakowski: *one option would be
library_name/OMS_name?sym=symbol_name
*[09:24] **TillMossakowski: *the problem is that this only works for
newly created DOL ontologies, not for existing ontologies
*[09:24] **ChristophLange: *curl -sH 'Accept: application/rdf+xml' -L
'http://xmlns.com/foaf/0.1/Person'   (you see the problem: although you
have selected only one symbol, the whole document is downloaded)
*[09:24] **TillMossakowski: *Existing ontologies use
library_name/OMS_name#symbol_name or library_name/OMS_name/symbol_name
*[09:26] **TillMossakowski: *or better: OMS_name#symbol_name or
OMS_name/symbol_name
*[09:26] **TillMossakowski: *in MMT: OMS_name?symbol_name
*[09:27] **TillMossakowski: *we do not want to encourage URNs, because
we want to encourage a linked-data style, which is incomatible with URNs
*[09:28] **TillMossakowski: *that said, we do not want to forbid URNs
*[09:33] **TillMossakowski: *we suggest that the user can specify the
separator, and it defaults to #
*[09:34] **TillMossakowski: *more precisely, we need a library-OMS
separator (defaulting to /) and an OMS-symbol separator (defaulting to #).
*[09:36] **TillMossakowski: *Tara: it is quite common that an the IRI of
an OWL ontology ends with a #
*[09:37] **TillMossakowski: *in this case, the user could let the second
separator be the empty string
*[09:39] **TillMossakowski: *this means that separators can be arbitrary
strings
*[09:41] **TillMossakowski: *the problem with # is that the http server
only gets the OMS IRI, while the http client has to interpret the #
*[09:45] **TillMossakowski: *so we suggest to let both separators
default to / (even if this causes certain ambiguitites, because / is now
1. a separator between folders in the library IRI, 2. a separator
between library IRI and OMS name, and 3. a separator between OMS name
and symbol name. However, this can be disambiguated dynamically)
*[09:46] *List of attendees: ChristophLange, TaraAthan, TerryLongstreth,
TillMossakowski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.ovgu.de/pipermail/ontoiop-forum/attachments/20140924/427abfe1/attachment.html>