Word Embedding in Natural language processing