Whether a person is one of the high or low performers – i.e., above average or below average performance and success – usually only becomes apparent after a long time in everyday work or study.
A Russian scientist has shortened this path by trying to understand on the basis of social media posts whether the respective originator is a high or low performer, as the science magazine “Eurekalert” reported. An artificial intelligence specially created for this purpose examines the contributions for certain words and their length as well as characters and symbols and uses the results to make a statement about the academic performance of the author.
The study was carried out by the Russian natural scientist Ivan Smirnov, senior research fellow at the Laboratory for Computational Social Sciences at the Pedagogical Institute of the Higher School of Economics in Moscow, and published on the freely accessible platform “EPJ Data Science”.
Language suggests education and success
In order to be able to establish a connection between the academic career and the contributions on social media, Smirnov used data on around 4,000 former PISA participants and students from all over Russia as well as their public contributions on the Russian social media platform VK.
Using mathematical text analysis, over 7 million social media posts were examined and certain linguistic features were worked out, which should indicate a higher or lower level of education of the author. These characteristics or individual words were then classified in a rating system in order to clarify the level of education of the users.
As a result, it can be determined that the use of emojis, exclamation marks and capital letters stands for lower academic education and success. Longer words and posts in general, as well as a broader vocabulary and high information content, on the other hand, would be indicative of a higher academic education and the success of the author.
In addition, words from certain word families or fields are associated with more education and greater academic success than others.
Words associated with better academic performance include, but are not limited to: words and families of words around books and literature (author names, words from the publishing world), physics terms, words relating to thought processes and cognitive performance.
Words associated with poor academic achievement included misspelled words, computer game titles, military and service terms, and words related to automobiles and car accidents.
In the future, income could also be predictable
Since it is an artificial intelligence, the program learns by itself over time to classify and evaluate words. If the AI realizes that the names of literary people are more likely to be mentioned by high performers, the program will automatically rate contributions with corresponding mentions better, even if the name of the character itself may have been completely unknown.
In addition, according to the science magazine, the program also works with data sets from other sites (such as Twitter), which proves that the model could ultimately also be used in completely different contexts. For example, instead of academic performance, it is conceivable to make statements about a person’s future income or mental state.