27th April 2015, 06:37 PM
barkingdigger Wrote:Since computers can't think, their algorithm-generated predictions are only as good as the data they have to chew on!
Hence the 'big data' thing - it has to be really big data, you have to look at whole populations over years and such like, but then it starts turning up stuff like people get mugged on particular American streets but only when the temperature gets over 25 degrees and the like. Archaeology doesn't have a large enough dataset to work with, quite apart from problems with different ways of recording stuff.
Having said all that, an effort like the Roman Rural Settlement project using a much bigger dataset than usual seems to have generated some new and interesting (and sometimes unexpected) stuff, so more seems to be better in a general way, but it's not going to tell you that there are going to be Neolithic pits in a particular field for instance.