Print

Print


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Am 26.07.2014 02:17, schrieb Tennant,Roy:
> If I ran into ISBD in a dark alley I would take it out with my pocket
> knife and do us all a big favor. Using punctuation to make a statement is
> so last century. Computers should not have to do parsing of text strings
> to know what you are trying to tell us. Is there a bracket? Yes? No? Is
> there a period? A comma? A slash? And the punctuation is there whether I,
> in my application, care to display it that way or not. ISBD was made for
> catalog cards and it's long past time for us to realize that those days
> are long over. Display needs to be completely divorced from data.
> Semantics should not be embedded into text strings.

True. As long as we talk about machine representations of something.
When it comes to humans consuming that data the picture may be different:

An ISBD description is serial text, telling me a story. Different
sentences in that story may belong to distinct semantic aspects,
but usually the human mind is able to recognize this implicitly.
"Thomas Mann's Zauberberg" or "Der Zauberberg / by Th[omas] Mann"
usually give sufficient information without bothering us with
labels like "title proper" or "statement of responsibility" which
I'm forced to understand first and match against my internal
concepts secondly, thirdly concluding that they are compatible
enough or don't fully understand and don't care anyway.
(I'm often amazed by myself ;-) how selective I can divide
regular mail from junk mail in my inbox just by looking at
the purported names of the senders).

ISBD display helps for foreign languages or scripts, its unobtrusive
markup just gives slight hints of changing or refined semantics.
I may pick them up or ignore them. (The deficiency of paper cards
of course is, that you cannot pick up these hints unless you
know the rules, hypertext could provide the additional layer(s!)
to selectively follow semantic hints in the display.

With XML documents there was a distinction between data centric
and document centric approaches, often characterized by the
permission of "mixed content". Traditional bibliographic
records somehow redundantly follow both approaches, i.e. you
have "data" in MARC 100 and 700 and the same facts recorded again
as parts of the text in 245$c. Especially in cases where the
redundancy is not very high, i.e. when the form recorded in
the statement of responsibility grossly deviates from the form
given in the heading one would wish for additional markup
linking the substring in the SoR with the heading or - like TEI does -
embedding heading information in markup distinguishing the name
in the SoR.

Now RDF (with string data types) enforces a strict data-centric
view on our bibliographic situation which even in circumstances
we usually consider as "pure data" fails to provide appropriate
descriptions.

My pet example are names of persons, there are forms like "Thomas
Mann" and "Mao Zedong" used in the real world (or 245$c or in
wikipedia) and forms like "Mann, Thomas" and "Mao, Zedong" as
seen in headings (also phone directories &c.). There exist
cultural concepts of "given name" and "family name", and also
of "first name" and "last name", depending on culture the real
world forms have one preceding the other or the other way round.
The strings "Thomas Mann" and "Mao Zedong" /have/ an underlying
structure, and our access point forms introduce an additional
delimiter ", " to make this more explicit (BTW its not ISBD to
blame here).

RDF descriptions like < http://d-nb.info/gnd/118577425/about/rdf >
are trying to exploit the internal structure, but in my opinion
gloriously fail with that: Remember that any variant form of
an access point can be analyzed with respect to parts of names,
and therefore fragments like

<gndo:variantNameForThePerson>Mao Czė-dun</gndo:variantNameForThePerson>
...
<gndo:variantNameEntityForThePerson rdf:parseType="Resource">
  <gndo:personalName>Mao
Czė-dun</gndo:personalName></gndo:variantNameEntityForThePerson>

and

<gndo:variantNameForThePerson>Maŭ, Zedong</gndo:variantNameForThePerson>
...
<gndo:variantNameEntityForThePerson rdf:parseType="Resource">
  <gndo:forename>Zedong</gndo:forename>
  <gndo:surname>Maŭ</gndo:surname>
</gndo:variantNameEntityForThePerson>

the (1:1) connection between the distinct "variantNameEntityForThePerson"
and their respective "variantNameForThePerson" is completely unexpresssed.

Also <gndo:forename>Zedong</gndo:forename><gndo:surname>Maŭ</gndo:surname>
does not imply any order or prescribes a delimiter, therefore a
variantNameForThePerson cannot be constructed from a
variantNameEntityForThePerson (at least without additional rules to be
known by applications).

A satisfying solution, i.e. making internal structure of RDF elements
explicit without resorting to delimited strings, would employ the rdf
parseType "Literal" (rdf:XMLLiteral) and mixed content:

<gndo:variantRWNameForThePerson>
  <my:name rdf:parseType="Literal" xmlns:my="http://mynames.example.org">
    <my:familyName>Maŭ</my:familyName> <my:givenName>Zedong</my:givenName>
  </my:name>
</gndo:variantRWNameForThePerson>

<gndo:variantRWNameForThePerson>
  <my:name rdf:parseType="Literal" xmlns:my="http://mynames.example.org">
    <my:lastName>Mann</my:lastName>, <my:firstName>Thomas</my:firstName>
  </my:name>
</gndo:variantRWNameForThePerson>

Thus "native" XML is used to express the internal structure of some RDF
elements.

Its no surprise that analogous constructions would be more complicated
when it comes to transcribed elements like statements of responsibility
rather than lists of variant name forms in authority data:

If we would edit an ancient catalogue, it would be possible to
express in TEI:

... / by <tei:name type="person">
           <tei:choice>
             <tei:orig>Th. Mann</tei:orig>
             <tei:reg>Mann, Thomas</tei:reg>
           </tei:choice></tei:name>

or - using the TEI extension module for name &c.:

... / by <tei:persName><tei:forename>
  <tei:choice><tei:sic>Th[omas]</tei:sic>
              <tei:abbr>Th.</tei:abbr>
              <tei:expan>Thomas</tei:expan></tei:choice>
</tei:forename> <tei:surname>Mann</tei:surname></tei:persName>

This persName elements can now be linked to centralized
tei:person Elements within the document, which serve to
collect biographical data for identification and reference.
Either tei:persName or tei:person may also be linked to
external resources like authority files.


Since more modern calaoguing codes are trying to tamper less
with the original when transcribing content found on the
resource, I see the demand for inlined annotations increase:
* This "Th." stands for "Thomas", not "Theodor"
* This "Th. Mann" is the form of the name given for the person
  identified as the author of the resource

Thus like in the TEI scenario a bf:responsibilityStatement
should allow internal structure (i.e. arbitrary XML) able
to insulate and annotate forms of names, and furthermore
should link this name to the person referenced by the name -
preferably to resource-specific statements about this person.
The latter would be one of our bf:creator or bf:contributor
statements, thus in contrast to the "regular" XML situation
we have two additional obstacles:

1. Our evidence fragment "Th. Mann" resides on manifestation level,
   but creators and contributors are work- or expression
   level elements: Therefore we cannot even think of
   embedding the bf:creator statement within the arbitrary
   XML within the bf:responsibilityStatement

2. "Linking" in RDF (here: from something to some aspect of
   the description noted elsewhere) requires turning this aspect
   into a resource, i.e. reification.

To conclude: I have no idea at all how the goals of more
faithful transcription of information found "on" resources
(without loss of functionality, i.e. explanation and linking)
and expressing everything in RDF can be simultaneously obtained.
Reducing all literals to strings makes things simpler, but
especially in that situation I see no reason to riducule ISBD
which does perform a good job in presenting several layers
of information within strings without out-of-band mechanisms.

viele Gruesse
Thomas Berger


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iJwEAQECAAYFAlPU12sACgkQYhMlmJ6W47OZxAP9HqbVKjgC26LpUyDFKl64Gx5U
eRk9Oxf8sg8sBbMmGSMTG1IsqI/dHSKPa2FW/0hLubAHBkiEaXCg+psNXA8cHhZq
exxYz/M7MrrioRckO4nVFhZIBasucwbj8N3Yk3DxOOMF9ASzCKKMbY2Knovsktqx
9N0hBN/ZL2bioMtCrW8=
=I6bT
-----END PGP SIGNATURE-----