Russian scientists have created an AI system that predicts a student's academic performance with approximately 88% accuracy based on their social media activity. This was reported by the press service of the Higher School of Economics (HSE).
Some results surprised us. For example, that students interested in art or travel show excellent academic performance. These hobbies do not interfere with studies. On the contrary, they seem to help them study better. And active engagement with communities related to part-time jobs turned out to be a marker of low academic performance, which is understandable.
In a study aimed at developing methods to help universities identify the most and least successful students, as well as choose strategies to improve the quality of education for both groups of students, researchers analyzed the open profiles of 4,400 students from Tomsk State University and compared the results with their grades.
The researchers studied data from social networks using the BERT neural network. It allowed them to determine what students were discussing in their groups and communities and divide their interests into several main categories. These data were then analyzed using several classification models. As a result, it was possible to identify a connection between social media subscriptions and student performance.
Students with high grades often become members of communities dedicated to science and education. They prefer more complex texts and show interest in in-depth discussions and information analysis. At the same time, students with low grades most often join entertainment groups that focus on humor, memes, music, and video games. In such communities, more attention is paid to negative emotions, and there is less information.
Further research has shown that, taking these aspects into account, artificial intelligence systems can predict student performance with high accuracy, namely in 88% of cases, based on open data from social networks.
As Dmitry Ignatov, Associate Professor at HSE University, emphasized, this approach will allow higher education institutions to identify promising students and adapt curricula to specific groups of students. Employers, in turn, will be able to use this method to select suitable candidates for vacant positions. The scientist also noted that when providing access to their data on social networks, it is important to consider this aspect.
Earlier, the neural network successfully passed the entrance exam to the Gnessin State Musical College. GigaChat coped with the tasks that applicants solve at the colloquium.
Read more on the topic:
AI-Powered Program for Creating Serial Killer Profiles Developed in Russia
Doctors Accelerated the Launch of Clinical Trials by Tens of Times with Yandex's Neural Network
Three Main Mistakes That Prevent Maximizing the Potential of Neural Networks
