Open daniel-acuna opened 9 years ago
Sure, I will add the function to train tfidf
and Truncated SVD
.
Awesome. In this case, we need to keep the pipeline around for when we compute an author's "mean topic vector". Before, we were creating an imaginary abstract with all abstracts concatenated. Now, we won't need which can potentially be faster.
Transform each of the abstracts of the authors through the Tf-Idf -> truncated SVD and then average. Now, it puts all abstracts together in a big text and does the transformation.