imclab / topbox

Python 2 & 3 wrapper around the Stanford Topic Modeling Toolbox. Intended to be used for hassle-free supervised topic classification with Labeled Latent Dirichlet Allocation (L-LDA, LLDA, sLDA).

topbox

A small Python 3 wrapper around the Stanford Topic Modeling Toolbox (STMT) that makes working with L-LDA a bit easier; no need to leave the Python environment. More information on its workings can be found on my blog.

Setting up

Just download STMT and put it in the box directory. After, import topbox from wherever you left it.

Example

import topbox

stmt = topbox.STMT('bit_of_testing', epochs=10, mem=15000)


space = ['text text more text', 'things to do with text']
labels = ['label1 label2', 'label1 label3']

stmt.train(space, labels)


infer = ['this is a text', 'things with more text']
gs = ['label1 label2', 'label1 label3']

stmt.test(infer, gs)


from sklearn.metrics import average_precision_score

# array requires numpy and scipy
y_true, y_score = stmt.results(gs, array=True)

print(average_precision_score(y_true, y_score))