Describir: Is singular value decomposition useful for word similarity extraction?