I would like to inform you that there is a very similar approach to add information about classification data to the MARC format.
Driven by the Dewey experts at OCLC and at the Library of Congress, and by the German National Library, there will be a paper on the MARBI agenda for June this year. The goal is to add content designation to the MARC format for provenance information, or data about data.
The draft paper as it is now has two sections, one about the immediate need of adding subfields to MARC 082, 083 and 084; and a second section about a broader approach that covers every MARC field, by defining a new field in the 88X range, and using $8 to link the two fiields together inside the MARC record.
Data about data, or provenance information, may be split up into (at least) "Method of assignment" (basically in binary form, assigned by a machine, or not), and "Process of assignment" (by giving a URI, a process name, or some other description), and a "Confidence value", which is a numerical value. These three parts may each form one newly defined subfield in 082, 093, and 084. A subfield $q "Assigning agency" already exists in these fields. The new field 88X may contain them as well, plus additional elements, as date and time information, and an identifier leading to an authority record.
Regarding the timeline, we are still working on a complete draft, which then will have to be sent to the colleagues at NDMSO, and after some steps I expect the paper to be put onto the MARBI agenda and to be published some weeks from now.
I have talked about this process during a meeting of the MODS/MADS Editorail Committee yesterday. The Committee suggested to inform you and the whole community, and maybe wait what direction the discussion will take, and then see how this may work in MODS and MADS as well.
German National Library
IT / Office for Data Formats
D-60322 Frankfurt am Main
mailto:[log in to unmask]
***Reading. Listening. Understanding. A century of the German National Library***
Von: Metadata Object Description Schema List [mailto:[log in to unmask]] Im Auftrag von Saa¹ha Metsarantala
Gesendet: Freitag, 20. April 2012 17:07
An: [log in to unmask]
Betreff: [MODS] machine generated classification in MODS
Items are often classified according to different classification systems.
According to the MODS schema, we can put several classification elements
with different @authority attributes, like this:
This is very useful! Good!
Sometimes, each classification is achieved by a human. In other cases, a
classication can be automatically generated by a software translating from a
classification formerly done in another classification system. Such a
machine generated classification is often less accurate and / or less
granular than a classification done by a trained human.
I wonder whether storing an automatically generated classification can be
considered being within the scope of what MODS intends to accomplish and if
it is the case whether we could add a new @generator attribute on the
classification element. This would be an indication that the classification
is automatically generated and it could also be used to store which software
was used to achieve this translation. As a work-around, we could of course
use the @edition attribute similarly to
classification authority="ddc" edition="22 machine generated from lcc by
foo-software version 1.2"
but I consider it would be more easily parsable (through XSLT etc.) to write
classification authority="ddc" edition="22" generator="lcc2ddc-foo-software
or (maybe even better) through a URI
classification authority="ddc" edition="22"