A Study on Word Similarity using Context Vector Models