Translation. Region: Russian Federal
Source: State University Higher School of Economics – State University Higher School of Economics –
A team of Russian researchers, including scientists from the National Research University Higher School of Economics, used artificial intelligence to analyze 4.5 thousand student subscriptions to VK communities. It turned out that the algorithms can predict with high accuracy who is an excellent student and who has difficulties with their studies. The work was published in the journal IEEE Access.
A person leaves behind a digital trace – likes, photos, information about listening to music and clicking on links. Even the most careful people can learn a lot from their Internet activity. Some believe that you can not monitor your digital trace and that information on social networks does not affect your professional and personal life. For scientists, open data on the Internet is a subject of research.
A group of scientists from the National Research University Higher School of Economics, Skoltech, and Tomsk State University collected data on subscriptions of 4,445 students with open profiles to various VKontakte communities. Then, using NLP analyzers (natural language analysis methods), they classified the topics of the communities, assessed the complexity of the texts that students read, and the emotional tone of the content. For each student, the researchers created a digital profile, including preferences and interests. After that, the scientists used machine learning to find a relationship between online activity and academic success.
The researchers created an algorithm that predicts academic performance based on subscription analysis. In particular, students with high grades are more likely to be subscribed to communities related to science and education topics, where new technologies are discussed and analytical articles are published. Excellent students read more complex texts and show greater interest in discussions and deep analysis of information.
Low-performing students were more likely to subscribe to entertainment communities that focused on humor, memes, music, and video games. The content of these communities was more likely to display negative emotions and was also less informative than that of higher-performing students.
“Some of the results surprised us. For example, that students who are interested in art or traveling show excellent academic performance. These hobbies do not interfere with their studies. On the contrary, they seem to help them study better. And active interaction with communities related to part-time work is a marker of low academic performance, which is understandable,” comments Sergey Gorshkov, a postgraduate student. Department of Data Analysis and Artificial Intelligence Faculty of Computer Science National Research University Higher School of Economics.
Educational organizations can use this approach to identify talented applicants and tailor curricula to specific groups. In addition, subscription analysis can help employers in recruiting, allowing them to find candidates with a high expected level of analytical skills.
“This study once again reminds us of the need for digital hygiene. For example, in agreements on opening an account at a bank or with a mobile operator, you can see that you give permission to use some information from a social network account linked to your phone number. This can then be used to create a digital profile. Whether you want this is up to you,” says Dmitry Ignatov, head of the Scientific and Educational Laboratory of Models and Methods of Computational Pragmatics at the Faculty of Computer Science at the National Research University Higher School of Economics.
Please note: This information is raw content directly from the source of the information. It is exactly what the source states and does not reflect the position of MIL-OSI or its clients.