ISSUE 186 [Re: SKOS comment: Last Call Working Draft]

Dear Michael,

Dear Michael,

Thanks again for your comment below (from [1]), which we have
filed as ISSUE 186 [2].

[..]

 > We also see potential problems in deriving the mapping
 > relations skos:broadMatch and skos:narrowMatch from
 > skos:broader and skos:narrower. In ISO standard and current
 > practices many multilingual thesauri did not use broader or
 > narrower to indicate the mapping relations. SKOS should
 > revisit those standards and follow the current standards'
 > development to make sure SKOS is consistent in representing
 > the indicators used by standards (and the thesauri following
 > those standards) for so many years.

Quating from Alistair Miles' response [3]:
"The SKOS mapping properties have their roots in ISO 5964 and 
have been informed more recently by BS 8723 part 4. I don't have 
ISO 5964 to hand so I may not quote precisely from it here, 
hopefully others can correct me if I make any glaring errors. 
ISO 5964 introduced the notions of exact, inexact and partial
correspondance between thesaurus descriptors. These provided
inspiration for the SWAD-Europe report on inter-thesaurus 
mapping, which first described the use of the SKOS Mapping RDF 
schema [4].

Note in particular that the "partial" correspondance as 
described in ISO 5964 indicated that the meaning of one 
descriptor *either* subsumes *or* is subsumed by the meaning of 
the other. Hence [4] refined the notion of a partial mapping to 
provide broad and narrow mapping properties, which are clearly 
more useful than the ambiguous "partial". We felt this was 
consistent with the intention of ISO 5964 (see also note [5]).

BS 8723 part 4 ("interoperability between vocabularies") 
provides a clear (IMO) discussion of mapping between extant 
vocabularies. It illustrates the use of standard hierarchical 
and associative relationships (BT, NT and RT), in addition to an 
equivalence (EQ) relationship, to assert mappings between 
vocabularies, in what they call "differentiated mappings". These 
directly correspond to the skos:broadMatch, skos:narrowMatch, 
skos:relatedMatch and skos:exactMatch properties. (See also note 
[6].)

We realise that "undifferentiated mappings" (where the nature of 
the correspondance is not specified) may represent the majority 
of real world mapping data. However, "differentiated mappings" 
are also an important resource, and are being constructed at 
scale e.g. by FAO.

Hence the current design for SKOS is based on a perceived 
consensus for mapping between vocabularies, which is to ground 
the different types of mapping relationship in the notions of 
hierarchical and associative relationships, and we believe that 
this consensus is consistent with existing standards."

 > In addition, when mapping systems that are structurally
 > heterogeneous (e.g., classification systems and thesauri), the
 > links established through mappings have no hierarchical
 > implications at all.
 >
 > Currently, skos:broader is used both for the hierarchical
 > relationship between classes as well as between concepts.
 > Mapping relations that are subproperties of
 > skos:broader/skos:narrower are not able to sufficiently
 > support interoperability between structurally heterogeneous
 > systems.

We understand your point. SKOS mapping relations cannot solve
the heterogeneity of vocabularies and it is not possible to
prevent wrong usage of the mapping relations. However, we think
that the mapping relations do provide an important mechanism.
Also, people can use, next to the broader/narrower, other
mapping relations such as closeMatch and relatedMatch, which
might be more suitable in heterogeneous cases.

We propose to add a note to the current text to clarify this point.

 > In addition, many different indicators of degree of mapping
 > have been used in integrated vocabularies, e.g., major
 > mapping, minor mapping, alternative mapping, and overlapping.
 > These may make the mapping properties even more complicated.
 > The solution here might again be to extend mapping properties.

Our SKOS design rationale [5] is:

[[
    "The notion of a Knowledge Organisation System encompasses a
     wide range of artefacts. There is thus a danger of
     overcommitment in the SKOS schema, which could preclude the
     use of SKOS for a particular application. In order to
     alleviate this, in situations where there is doubt about the
     inclusion of a formal constraint (e.g., seediscussion
     on skos:hasTopConcept), the constraint has not been stated
     formally. In such cases, usage conventions may be suggested,
     or specialisations of the SKOS vocabulary may be used in
     order to enforce constraints (see the SKOS Primer)."
]]

So, we agree that extending the mapping properties might very
well be a good idea, but we prefer to leave this to developers.
See also the section in the SKOS primer on extension mechanisms
[6].


We hope you live with this response.

Regards,
Guus Schreiber

[1]
http://lists.w3.org/Archives/Public/public-swd-wg/2008Oct/0061.html
[2] http://www.w3.org/2006/07/SWD/track/issues/186
[3] 
http://lists.w3.org/Archives/Public/public-esw-thes/2008Oct/0041.html
[4] http://www.w3.org/2001/sw/Europe/reports/thes/8.4/
[5] http://www.w3.org/TR/2008/WD-skos-reference-20080829/#rationale
[6] http://www.w3.org/TR/skos-primer/#secskosspecialization

[7] IMO ISO 5964 requires careful interpretation. From previous
readings, I understood that ISO 5964 is primarily aimed at 
describing the *process* of constructing a single multilingual 
thesaurus, *not* on mapping between extant monolingual thesauri 
in either the same or different languages. The notions of 
"exact", "inexact" and "partial" are used to describe the types 
of correspondance that can be encountered between different 
language components *during the process of constructing a 
multilingual thesaurus*, with the implication being that 
anything other than an exact correspondance must usually be 
mored closely aligned before the thesaurus is finally published.

[8] Although the main body of BS 8723-4 discusses mapping 
between vocabularies (sections 5-8), where the assumption is 
that modifications to each vocabulary cannot be made to improve 
the alignment, BS8723-5 also discusses the process of 
constructing a single multilingual thesaurus (section 9), where 
changes can be made to each language component to improve the 
overall alignment of the thesaurus. IMO section 9, whilst 
valuable, is out of place in BS8723-4, because the process of 
constructing a single multilingual thesaurus (where language 
components can be modified to improve alignment) is different 
from the process of mapping between extant thesauri (where 
mappings have to describe aligment as-is), and would be better 
treated in a separate document.