how can I extract document-topic matrix from LDA model and use it as input features an svm classifier? I am using gensim for implementation


I've done this before in Gensim, hopefully it will help:

train_vecs = []
for i in range(len(your_training_examples)):
    top_topics = lda_train.get_document_topics(train_corpus[i], minimum_probability=0.0)
    topic_vec = [top_topics[i][1] for i in range(20)]

The above would give the top 20 topics for every document. 'train_corpus' is the result of doing something like this in Gensim once you have a bigram object from the 'Phrases' Gensim model class:

train_corpus = [id2word.doc2bow(text) for text in bigram]

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.