This repository contains code for preprocessing twitter data labeled as straight or gay, and then run a Metropolis-Hastings random walk. Requires:
- SNAP (snap.stanford.edu)
- Rochester twitter graph and user list
- Several python libraries, including networkx and pqdict.
This repo contains the necessary labeled data. To run, clone this repo into the examples directory under snap. Then edit the Makefile with the the location of any python libraries.
You need the following files.
- 02-2015_master_user_list.json
- 02-2015_social_graph.json
These are located, for instance, on
- cycle1.cs.rochester.edu
in
- /p/twitter4/rochester/02-2015
To build the larger neighborhood graph, type
export DATE=02-2015; make ${DATE}_twitter_graph.pkl
To build the input to the MHRW, type
export DATE=02-2015; make ${DATE}_gaydar_snap.graph
To compile the mhrw, type
make
Here is an example of running the mhrw.
./mhrw -file:gaydar_snap.graph -balance:.5 -flip:0 -num_swaps:100 -grad:2500 -gayweight:10 -labweight:10 -greedy:0
The arguments are explained in mhrw.cpp