LDA_SCVB

Latent Dirichlet Allocation using Stochastic Collapsed Variational Bayes (Foulds et al 2013).
paper here

Usage

This model allows you to specify batch or online processing. Future updates will create an update function.

    //LDA(iterations, numDocs, numTopics)
    LDA lda = new LDA(30, 1500, 20);
    lda.processing("batch");

    lda.trainNIPS("./text/docword.nips.txt","./Vocab/vocab.nips.txt");
    
    ArrayList<HashMap<String,Double>> termTopics = lda.termTopicProbs(10);
    
    Print.printTopicWords(termTopics);

Results

Reading Files...
LDA updates:
Iteration: 1...... Done.
Iteration: 2...... Done.
Iteration: 3...... Done.
.
.
.
Topic: 0
university : 0.040116595764092516
mean : 0.03885503115118312
error : 0.03702257269693845