Code for fitting a topic and point of view (POV) model of collective intelligence processes, using parallel Gibbs sampling. Also includes synthetic data generation for testing and evaluation.

To compile and run:

make
bash test_synth.sh 50 500

You will need the GNU Scientific Library (GSL) headers installed to compile everything. See comments in test_synth.sh for a high-level overview of the inference process.

Files associated with executables (descriptions at the top of each file):

• basicstats.c
• check_indexes.c
• compare_users.c
• inference.c
• initialize.c
• page_user_stats.c
• readout.c
• set_assignments.c
• store_revisions.c
• verify_mmaps.c
• word_probability.c

Shared functions, struct definitions (descriptions in header files):

• index.h
• comparisons.h / comparisons.c
• parse_mmaps.h / parse_mmaps.c
• probability.h / probability.c
• sample.h / sample.c

Python helper scripts (in bin/):

• compare_clusterings.py
• synth.py
• topic_pov_to_clusters.py

Bash to glue everything together:

• test_synth.sh
• initialize.sh