Abstract:

Nowadays, classrooms are crowded with large numbers of pupils. As stated by the Organisation for Economic Co-operation and Development on Europe [1], in OECD countries there are more than 21 pupils per class, on average, in primary education. Nonetheless, the numbers tend to be higher in other G20 countries. Total class sizes range from over 29 in Chile and China to almost half that number in Luxembourg and the Russian Federation. In lower secondary education, the average class in OECD countries has about 23 students and 26.7 in the UK [2].
Large class sizes mean that teachers have less time to invest in each student, which ultimately leads to lower academic performance.

Some countries have positions designed to cater for specific types of students. For instance, in Spain there is the figure of the Therapeutic Pedagogy Teacher, who helps students with physical and mental problems; and the figure of the Hearing and Language Teacher, who helps students with language-related tasks. This approach has proved positive for the learning issue, so the conclusion is that having a lower teacher to student ratio is highly beneficial.

To further assist in this matter, we propose Qasor: a question-answering social robot in class. Our proposal is based on the Pepper platform, which is a humanoid robot. It has a range of sensors such as touch surfaces, cameras, microphones and speakers. We have implemented in it a machine learning-based question-answering engine that is able to take as input a plain text and answer questions about it. Once the text is loaded by the teacher at the beginning of each class, students can ask questions in natural speech and Quasor will answer them in natural speech as well. Our approach shows high accuracy in the question-answering task and will improve the class participation and learning process of the students.

References:
[1] EDUCATION AT A GLANCE 2012: HIGHLIGHTS. Accessed 17 September, 2022. Retrieved from https://www.oecd-ilibrary.org/docserver/eag_highlights-2012-25-en.pdf?expires=1665998127&id=id&accname=guest&checksum=4630401C2621CCDC31A188059F595D7F
[2] Schools, pupils and their characteristics. Accessed 17 September, 2022. Retrieved from https://explore-education-statistics.service.gov.uk/find-statistics/school-pupils-and-their-characteristics

Keywords:

Question-answering, social robot, machine-learning.