This paper investigates the approximation problem of large-scale Boolean networks (BNs). The whole network is partitioned into several blocks and the input-output representation of each block is obtained via the observed data. By analyzing the simplified network composed of the input-output representation of each block, the properties of the original network can be obtained. The relations of topological structure and finite-time stability between the original system and the approximated system are discussed. Several illustrative examples are given to demonstrate the obtained results about the approximation of large-scale BNs.