Scaling-Up Inference in Markov Logic