An Unsupervised Language Independent Method of Name Discrimination Using Second Order Co-occurrence Features