SESSION REPORTS FROM CCUC/ 5 autaoring generated-question "pratoty?as" has already been imporvei bayoa l that described in the published piper, , Although thare are pcobaely othe r systems that exceed "IPSIM's" abilities in any singLe_ function,I im ne t aware of any which contain such a powerful c3mbinitiea of item seLection an d generation . Pete Olympia from the Washington racheical Institute described a refreshingly innovative approach to tea computer generation of questions . It-am generation oriinariLy consists 3! either filling in randomly chose n numeric values or selecting tee aptiois for multiple-choice questions from lirgar set . the algorithm used here is basal upon banking questio n components and assembling than in a :imam fashion (within limits) t o produce logical guestions . Of somewhat lass interest (because of it s widespread use), the system is also Sale to generate items by choosin g The technique of assembling text components is a namaric data at random . novel approach to item generation which one hopes will attract interes t among others, especially those with a bickgraand in syntactic tax': manipulation . It also suggests stubs determine the conditions unde r which the generated items fall into i
/lp/association-for-computing-machinery/ccuc-6-chemistry-lqB6y0uMvZ