Original mapping by similarity release

Transfer learning with word embeddings-based similarity. Mapping across domains using Soft Cosine similarity, Euclidean Distance, and Word Mover's Distance. Pre-processing by segmenting predicates into their components words, verbs into their regular form, and replacing shortened words with their full form.