Abstract:

In this paper we will present PERTUS – PERsonalized TUtoring System – whose main goal is to provide each student distinct assignments and evaluation tests to promote a personalized learning experience. With PERTUS, distinct datasets and evaluation tests/exams can be created for each student. It is a program fully developed in Python whose main goal is to give each student a personalized problem instead of the traditional approach, which is to provide the same problem to all students. This personalization became especially important during the pandemic period when students were at home and the motivation to cheat evaluation instruments was high. To guarantee that every student receives a different problem, but with controlled complexity, random datasets are generated and tested until an acceptable one is found. Only then it is sent to the student. Otherwise, new datasets are generated until an acceptable one is found. The software also provides the tutor all the correct answers for each student’s problem allowing an easy and effective support during practical classes. When students finish their assignments, they fill a template received together with the assignment and return it to the course email address. Each answer is automatically analysed by the software and a grade is returned to the student with a report containing the detected deficiencies (if any). All communications between students and PERTUS are based on email messaging, guarantying the registration of all messages and allowing asynchronous activity. The system was applied on several courses with 150+ students covering from Artificial Intelligence to Image and Sensorial Processing but it can applied on any other area where problems are mainly numerical. Another important concern was the loss of trustworthiness in online assessments. While the majority of teachers opt for the imposition of hard time constraints to hinder the chances of students copying from each other, PERTUS generates different evaluation tests for each student, making copy almost unfeasible (especially if used together with the proctoring system available on Moodle). Once that each student will have a different resolution the system automatically generates the correct values for each student and intermediary values to help the evaluator during the evaluation.

PERTUS proved to be efficient in all situations. During practical classes, all the data for the assignments is produced in real-time and personalized to each student so students can develop their projects on a tailored way. The teacher has all the expected results available to assist the students during the assignments development and the automatic evaluation (with teacher supervision that gives the students a feeling of humanity that must not be neglected) is very welcome by the teachers (for obvious reasons) and by the students due to feeling of fairness. The generation of personalized evaluation elements was also a success because students were able to take their online theoretical evaluation exactly as they were with in-premises tests and it was very well received by them.
Therefore, we can say that PERTUS allows an innovative, friendly teaching/learning environment that will remain after the pandemics to produce a more enjoyable teaching and learning experience.

Keywords:

Personalized assessment, Personalized teaching, teaching technology.