Automatic test paper generation is a NP hard problem while restrictions exist.In this paper,an application of the PBIL(Population-Based Incremental Learning) algorithm to this problem is studied for the first time.The relationships between the optimal function and restrictions are emphasized in this paper.It is proved that the PBIL algorithm is a suitable method in solving this problem,and it is flexible to attach restrictions and is stable in the process.Also,entropy is used here to estimate the degree of ...