A multi-year research project at Dependable Systems Lab is aiming at collecting command-line data from *NIX users to authenticate users by their command usage. However, there are many privacy issues with sharing such data, which is why no one has done this before. I wrote a GUI in Perl/Tk called "Monalog Sanitizer" that allows users to go through their own logs and sanitize sensitive information by simply selecting it and pressing a button. Users can also use literal or regular expression searches over a large number of their log files. The users' sanitizations are stored separately from the actual log files, and can be applied to the logs to create sanitized data sets that can be shared with us (the experimenters) without compromising privacy. A side benefit is that the different levels of sanitization allow us to run controlled experiments to determine the negative effect that this witholding of information has on the "authenticating power" of their command-line data. The work has been presented at IEEE Dependable Systems and Networks conference. Click on the thumbnail on the left to see a brief set of slides with illustrated screenshots.
Please ask if you wish to see more of my Perl work. I have written many data crunchers and parsers in Perl.