Peter Iles wrote:
> Simple transcription errors are easily corrected, once you spot
> them, but other problems are more obdurate. One such is 'Fuzzy Space' i.e.
> what you do with sites that are insufficiently well spatially referenced,
> e.g. Mr Smith reported a prehistoric burial mound 'near Anytown'. If you
> accept this as a valid site - Mr Smith may have provided sufficient
> information to show that it wasn't made up - then how do you plot it on the
> SMR?
I don't know about ArcView, but normal RDBMS such as Access can fairly easily be
made to contain fields of 'fuzzy' data; it should be possible to interlink these
with a GIS and generate probabilistic maps. In Peter's example, the record for
the burial mound would have a point coordinate (at Anytown centre) and a
probability function such that there is, say, a 66% probability that the mound
lies within 500 m from that point, 95% probability that it lies within 1000 m,
etc. Even though the probability of the mound being present in any particular
location is very small, for planning control purposes it would generate a 'heads
up' signal, which is the main thing. I know of a few experimental applications
of 'fuzzy' data fields being used in combination with GIS and am currently
writing a chapter (part of my thesis) on the proper structuring of a regional
archaeological RDBMS. I'll send the relevant snippets to our moderator for
distribution on this list.
Martijn van Leusen
PhD student, University of Groningen
|