Fast Katz and Commuters: Efficient Estimation of Social Relatedness in Large Networks