Artificial intelligence predicts a student’s academic performance based on his tweets

A group of experts trained artificial intelligence to predict how well a student would perform academically, just by looking at their tweets, with an accuracy of more than 93%, as a computer model was trained on thousands of test scores and a million posts on social media to distinguish between academic high achievers and those below them. Based on text features shared in posts.

According to the British newspaper “Daily Mail”, the AI-assisted technology has determined that students who discuss scientific and cultural topics, along with writing long posts and words, are likely to perform well.

While those who use an abundance of emojis, full capitalized words or phrases and vocabulary related to horoscopes and leadership tend to have lower grades in school.

The study was conducted by a team from the National Research University Graduate School of Economics, who used a prediction model that uses mathematical textual analysis that is able to categorize words, phrases, topics and other content in social media posts.

The experiment collected test scores from 2,468 students who participated in the International Student Assessment Program (PISA), A test system used to measure students’ performance in mathematics, science and reading.

Besides the test, the dataset contained more than 13,000 social media posts, and the results were compared with the average of the standardized state test.

A total of 1.9 billion words were trained on the system, with 2.5 million unique words, and the model went to work with the classification of textual features in the posts, and a negative relationship was found between the use of uppercase words, emojis and exclamation marks, and academic performance.