All in all, the file classifier uses relatively simple techniques to correctly guess the types of all of the files it has run across so far. It can be argued that the data set chosen to tune the classifier is not representative, but with some further adjustments the classifier should become a useful production tool.