The following workflow will generate sets of social features for Twitter users 0. download c++ boost libraries and add to "util" folder - have it such that "Social-Features/util/boost/tuple/tuple.hpp" is a valid path 1. ensure c code is compiled: % cd features % make clean % make 2. add Twitter json files to "collection" folder - have it such that "Social-Features/collection/<filename>.json" is one of the files - scripts expect files in the format the Twitter streaming API uses 3. make a directory "results" 4. execute the lines in fsets/fsets.txt in order while in "results" folder Note: You'll need twittercollect and simplejson python packages installed You'll also need the unix program sort installed - commands need to execute in that order - some commands will take a while to execute - some scripts (conv and prop) require large amounts of memory - memory needed depends on the size of the timeline - also a number of intermediate files will be created. some are large. - total size of intermediate files is less than the size of the collection