On Thu, Jul 26, 2012 at 3:17 AM, Henny van Schie
<[log in to unmask]> wrote:
Michele,
In the National Archives of the Netherlands we do track revisions to our
finding aids. We have two systems, a manual one with <revisiondesc>, and an
automated one in our XML database.
The content model of the element <revisiondesc> is:
<revisiondesc audience="internal">
<change>
<date calendar="gregorian" era="ce" normal="20101021">21-10-2010</date>
<item>
<num>2</num>
Free text, not standardized, describing the change
<name>name of the agent, sometimes a script</name>
</item>
</change>
</revisiondesc>
Obviously, not all changes are recorded.
Recorded are:
Mandatory:
- all accruals, expressed by one or more added <c# level=”file”>
elements (an element with level=”file” contains the description of a
deliverable unit: a folder, box, book, map, etc.);
- splitting up a deliverable unit into two or more new and smaller
units, expressed by one or more added <c# level=”file”> elements and the
removal of the original one;
- removal of deliverable units as a result of appraisal or transfer
of archival material to other archives;
Highly recommended (the processing archivist decides):
- addition of new texts in <p>, <list> or <table> in the descriptive
block elements (<bioghist>, <processinfo>, etc.), at any level;
- addition of new annexes (concordances, definition lists,
glossaries, bibliographies);
- addition of new schemes (genealogical tables, organogram charts);
- removal of elements, containing text;
- general facelift of a finding aid (rearrangement of descriptions,
conversion of the 19th century spelling into modern spelling);
- generic adaptations (conversion from DTD tot Schema).
Not recorded are:
- individual corrections (typing and marking errors);
- corrections, additions and removals of small parts of texts of
individual elements;
- changes as a result of synchronization with other systems;
- changes in rendering and presentation (i.e. <table> replaced by a
<list>).
The automated process of tracking revisions (versioning) is done by our EAD
Management System. This system keeps a copy of all changes in all EAD
instances in our dedicated XML database. It is possible to compare any
specific version of an EAD instance with any other version of that
instance. It is like comparing selected revisions on the 'view history'
page of a wiki.
The manual tracking method is for reasons of accountability: our successors
might want to know what we have done with the archival descriptions and the
archival material. The automated versioning is to prevent disasters: if you
have lost a big, manually into EAD converted finding aid, by accidentally
saving an empty or corrupted version of it, you know what I mean ;-).