Lists
Lists
Almada, Arlindo, Yu, Qicheng and Patel, Preeti (2023) Assisting students with the proactive chatbot based on their group of problems. In: Paris Conference on Education, 16-19 June 2023, Paris. (Submitted)
Universities face difficulty grouping and individualising the students’ assistance among thousands of students. This paper is founded on the author’s previous works related to student assistance. One of the author’s last research gaps while assisting students using the proactive chatbot was distinguishing between helping falling students and the other groups of students. In other words, assist students accordingly to their group of problems. This paper proposes an automated way to assist students according to the number and correlations of factors affecting students’ academic performance. It aims to automate the chatbot, especially in identifying and predicting groups of potential failure students, and proactively help them differently than other students. We presented the research context and predicted the students’ academic results to achieve this aim. This experiment was developed with a dataset from the Angola experiment at Universidade Católica de Angola collected in 2018. The study results showed a correlation between the variables and an outstanding accuracy of 0.935, and a precision of 0.76 was achieved. Then, the correlation between the variables and different students’ clusters was found. Those results allowed us to build seven distinct groups/clusters. With that knowledge, we integrated the predictive model and the discovered clusters into the proactive chatbot. It resulted in paying more attention to the groups where students presented more problems. Accordingly, in our local experiment, the proactive chatbot could change the level of extroversion by interacting more often with failing students. However, a real-life intervention is needed to validate our results.
 |
View Item |