evaluate lda output using classification metrics

[Hajar Zankadi]

I am working on a research paper and using 3 different algorithms including LDA.

I trained LDA using both bow and tfidf corpus, I have computed the coherence score for both of them but I am also interested in computing recall and precision scores in order to do an evaluation of the three other algorithms and then a model selection

i want to use scikit learn for that, but I need to have my documents labeled, right?

if yes, how can I auto-label my documents (could you recommend me please a python code for that )

thank you for your consideration