Cosine Similarity of Large Data Sets in NLP with TF-IDF and Distributed Computing