Geography Reference
In-Depth Information
GIS coverages. The
Source Documentation System
(
SDS
) contains detailed information on
the history of British censuses and their reports, and provides a unique identifier for each
table within those reports. However, the current subject is our
Data Documentation System
,
implementing the DDI.
13.3.1 DDI elements not implemented
Our implementation ignores large parts of the original Data Documentation Initiative
standard, which identify
Collections
and the
Studies
of which they consist. At one level this
is a quite obvious decision: the whole of the GB Historical GIS is a single data structure, and
our data table is a single dataset or 'Study', so if need be we could provide this high-level
documentation via a single block of static XML. From a data librarian's perspective, however,
this approach is more questionable because the contents of our data table incorporate the
results of a series of quite distinct projects, including much data computerized elsewhere
and donated to us.
We are not data librarians and in the Great Britain Historical GIS two other metadata
sub-systems record the provenance of the data: the SDS locates individual data values in
particular rows and columns within named tables within the census reports, while a smaller
'Thanks' system allows us to associate each individual data value with a different set of
'credits' identifying the different roles of a whole series of contributors. Our aim with
donated data is always to check back against the original statistical publication from which
the data were transcribed, and if necessary modify the data so it matches the transcription
we would have created, so distinguishing between donors and the original historical source
is essential. However, none of this is the job of our DDS, which is about recording what the
data
mean
, and giving them a new context supporting time series analysis.
We have also not directly implemented
Category Groups
, although they are implicit in the
mappings we create between categories, or
Nested nCubes
, although some such mechanism is
desirable in presenting very large nCubes such as the almost unclassified 1841 occupational
data discussed earlier. We have also, obviously, not included either time or geography as
DDI variables, but record those using the separate structures already described: British
administrative geographies are too fluid to capture using one single hierarchy of units, and
the US NHGIS does not treat geography as a simple variable (Southall, 2001).
13.3.2 Extensions to DDI: cellRefs and universes
The Location Map is one of the key innovations of the DDI Aggregate Data Extension,
recording where data values occupying a particular cell within an nCube will actually be
found. As originally implemented by the United States National Historical GIS, this meant
identifying locations within conventional flat or hierarchical files held separately as ASCII.
Our implementation is very different, holding everything in that single column; and we
neither know nor care what order data values are within that column. We therefore defined
a new concept, a 'cell reference' or
cellRef
, which is simply a unique identifier which appears
in both the data table and the Location Map. Our cellRefs in fact have a readable structure:
the first part, in upper case, usually identifies the nCube the cell is part of, and subsequent
Search WWH ::
Custom Search