Bnprec

Reference

Gopalan, P., Ruiz, F. J., Ranganath, R., & Blei, D. M. (2014). Bayesian Nonparametric Poisson Factorization for Recommendation Systems. In Proceedings of the Seventeenth International Conference on Artificial Intelligence and Statistics (pp. 275-283).

@inproceedings{gopalan2014bayesian, title={Bayesian Nonparametric Poisson Factorization for Recommendation Systems}, author={Gopalan, Prem and Ruiz, Francisco JR and Ranganath, Rajesh and Blei, David M}, booktitle={Proceedings of the Seventeenth International Conference on Artificial Intelligence and Statistics}, pages={275--283}, year={2014} }

Installation

Required libraries: gsl, gslblas, pthread

On Linux/Unix run

./configure make; make install

** ** Note: UNTESTED on Mac ** ** Linux is strongly recommended. On Mac, you may run into issues. If you absolutely need it to work on Mac, let me know. **

On Mac OS, the location of the required gsl, gslblas and pthread libraries may need to be specified:

./configure LDFLAGS="-L/opt/local/lib" CPPFLAGS="-I/opt/local/include" make; make install

The binary 'gaprec' will be installed in /usr/local/bin unless a different prefix is provided to configure. (See INSTALL.)

GAPREC: Gamma Poisson factorization based recommendation tool

gaprec [OPTIONS]

-dir <string> path to dataset directory with 3 files: train.tsv, test.tsv, validation.tsv (for examples, see example/movielens-1m)

-m <int> number of items -n <int> number of users -T <int> truncation level

-rfreq <int> assess convergence and compute other stats <int> number of iterations default: 10

-alpha set Gamma shape hyperparameter alpha (see paper) -C set Gamma scale hyperparameter c (see paper)

-label add a tag to the output directory

-gen-ranking generate ranking file to use in precision computation; see example

Example

(1) ../src/gaprec -dir ../example/movielens -n 6040 -m 3900 -T 100 -rfreq 10

This will write output in n6040-m3900-k100-batch-alpha1.1-scale1-vb

You can change the settings for the Gamma hyperparameters alpha and c (see paper) using the -alpha and the -C options.

To generate the ranking file (ranking.tsv) for precision computation, run the following:

(2) cd n6040-m3900-k100-batch-alpha1.1-scale1-vb

../../src/gaprec -dir ../../example/movielens -n 6040 -m 3900 -T 100 -rfreq 10 -gen-ranking

This will rank all y == 0 in training and the test.tsv pairs in decreasing order of their scores, along with the non-zero ratings from test.tsv.

The output is now in a new directory within the fit. Look for ranking.tsv.