It would be really interesting to see a visualization (maybe using PCA) of the LDA vectors for each document. The topics are not super convincing that the LDA approach worked well.
Other than that, this is a good intro to NLP and calculating document similarity. Well done!
I'm very interested in ways we can expand LDA (and other topic models) to retain more of the meaning of the documents, especially for small feedback (like reviews), such that a human could explore the results and find impactful, actionable data.
Using N-grams increases the dimensionality way too much.
I've thought about word-vectors as an option, but am unsure how similar terms could be grouped internally in LDA.
This kind of subject is probably worth a PHD thesis, honestly.
