Custom LDA
Custom LDA topic modeling template based on R language tm package. Unlike existing VW LDA model this module create toptopics dataset.
## Gibbs Sampling parameters ##
burnin <- 4000
iter <- 2000
thin <- 500
seed <-list(2003,5,63,100001,765)
nstart <- 5
best <- TRUE
## Inputs ports ##
dataset1 - Corpus with document body as **prepReview** and document name as **InternalId**
dataset2 - A dataset containing a single column and a single row - **nb_topics**.
We created a separate dataset for number of topics instead of creating a parameter because this makes it easier to launch an experiment from azure functions.
## Output ports ##
1. Top terms per topic
2. Top topics per document
3. Gamma distribution
The source code is available [here][1]
[1]: https://github.com/ajakupov/azureml_custom
