This is a mirror from the repository located on Illimine here. This only contains the matlab source files; it does not contain the data. The link on Illimine contains the data.

**** Illimine Copyright ****

University of Illinois at Urbana-Champaign, 2015

illimine.cs.illinois.edu

Additional Copyright

This package contains the source code and the dataset used in the following paper:

Chi Wang, Xueqing Liu, Yanglei Song, Jiawei Han. Scalable Moment-based Inference for Latent Dirichlet Allocation. ECMLPKDD, 2014.

If you use any contents in this package, please cite:

@inproceedings{wang14, title={Scalable Moment-based Inference for Latent Dirichlet Allocation}, author={Wang, Chi and Liu, Xueqing and Song, Yanglei and Han, Jiawei}, booktitle={ECMLPKDD}, pages={290-305}, year={2014}, }

Code explanation

(1). Input is in folder /Data. It contains two files: Data/test.corpus, which is document-word file (each line is in the format 'docID wordID') and Data/test.dict, which is vocabulary file (each line is in the format 'word')

(2). Output is in folder /Data. It contains two file:

 a) Data/test.stod.mat. It contains 3 parts:

 -- 1. alpha0, which is the summation of Dirichlet priors

 -- 2. alpha, which is the inferred Dirichlet prior of each topic

 -- 3. twmat, which is a k (number of topics) x V (vocabulary size) matrix,
 the i-th row being the i-th inferred topic distribution p(w|topic=i)

 b) Data/test.topic.mat. It contains 1 part, a k x 1 cell, the i-th cell
 being the topic representation of the i-th topic (which is the top 10
 words ordered by p(w|topic=i))

Data explanation

Data/csabstract contains CS abstracts used in the paper.

For AP news dataset, please visit http://trec.nist.gov/data/docs_eng.html.

**** For More Questions ****

Please contact illimine.cs.illinois.edu or Chi Wang ([email protected])