Using TF-IDF normalization and word embeddings to predict the most likely categories for books based on their titles.