rtm

Regularized Topic Models

Implements the regularized topic models in Newman, Bonilla and Buntine [1]

Authors:

David Newman (newman@uci.edu) Edwin V. Bonilla (edwinbonilla@gmail.com)

Requirements:

This is a self-contained package. You will need to compile the necessary mex files:

mex gibbsmex_lda.c mex gibbsmex_semi.c

Main Contents

1. regularized_lda.m : Learns an RTM model.
2. make_reg_matrix.m : Builds regularization matrices
3. run_regularized_lda.m: An example of how to run the algorithms

Additionally, we provide in a separate file rtm_data.zip the "climate" dataset used in [1] as a test example: a) Ndw.txt: File containing the corpus data (b) vocab.txt: File containing the vocabulary

References

[1] David Newman and Edwin V. Bonilla and Wray Buntine. Improving Topic Coherence with Regularized Topic Models. Advances in Neural Information Processing Systems 24: NIPS'2011