Abstract: To overcome the common trouble that the papers generated by many test paper bank systems are unsatisfactory to teachers‚the usability of the generated paper was analyzed‚and an improved model was proposed‚which adds in the usability constraints on“avoiding overlapping knowledge points”‚“evenly distributing the question types‚contents‚ and difficulty levels”and“avoiding the repeat of questions in test papers”．Accordingly‚to solve the model‚an intelligent algorithm was presented based on a heuristic index called satisfiability degree．The improved model and its algorithm were used to generate papers for several test paper banks published by the Higher Education Press．The results show that the model is widely applicable to science‚engineering and medicine test paper banks．Moreover‚it completely avoids overlapping knowledge points‚decreases the paper-generating error and the uneven degree of question distribution． 

Key words: test-item bank, general test paper-generating model, usability, intelligent paper-generation, satisfiablity degree