Automatic test correction

The DynDevice LMS eLearning platform is equipped with an Artificial Intelligence system for automatic correction of open-ended learning tests.

The auto-corrector allows pre-correction of tests with open-ended questions completed by users, reducing the time needed for manual correction of tests.

Once the test has been pre-corrected, the human operator can edit the results suggested by the AI and change the result in hundredths to be attributed to each answer or to the entire test.

How does the automatic corrector of open-ended tests work?

The mechanism behind this system is that each open-ended question is associated in advance with a series of concepts that must be contained in the answer for it to be considered correct.

An example? Let's assume that the first question of the test is "What are the duties of the person in charge?".

When devising the test, eight concepts corresponding to the eight obligations of the person in charge must be associated with this question. In this way, the Artificial Intelligence will be able to recognise whether all eight concepts or only some of them are expressed within a user's answer and, on the basis of the percentage weight given to each concept, a score will be assigned to the user's answer.

And so for all other answers in the test.

Once the AI pre-correction phase is over, the human operator (usually the teacher) will have to validate the scores given by the AI.

Easy, isn’t it?