Romanian Media Landscape in 7 Journalists’ Facebook Posts: A Chatgpt Sentiment Analysis

Abstract This article used OpenAI Chat GPT 3.5’s sentiment analysis capabilities to evaluate 595 Facebook posts written by seven renowned Romanian journalists. Our main objectives were to: 1) Assess the journalists’ use of Facebook regarding journalism, including its effectiveness in promoting quality articles, combating disinformation, and facilitating explanatory journalistic procedures. 2) Allow an unbiased AI system to analyze the tone of the journalists’ social media posts, as well as their attitudes and opinions on the primary issues affecting the media environment in Romania. 3) Determine the framework that embodies the Romanian society’s 2023 ‘zeitgeist’ from recurrent themes, categorized and ranked, gleaned from journalists’ posts. We extracted data from public Facebook posts using Apify. Our sample included posts uploaded within a timespan ranging from one month old to one year old. It is worth noting that the owners of the public accounts analyzed were consistently involved and active. We also had to be able to accommodate with time and financial constraints, as well as adhere to the limitations imposed on the use of the specified free tools. Therefore, the sample set consisted of an initial group of 12 journalists with varied socio-demographic origins. Out of this sample, some journalists were removed following subsequent considerations, resulting in a final sample group of 7 Romanian journalists. Outcomes The journalists engage in discussions or endorsements of their own work or the work of others, strive to provide reliable and accurate information to the public before others do, as well as analyze and expose false information and manipulative messaging. Aside from informative content, they share images featuring their personal lives, such as their children, feline companions, or acquaintances, just like the majority of the general users (although such posts are confined to the realm of personal space, as they rightfully should be). Even after accounting for false flagging, which ChatGPT rarely commits, the majority of the individual journalists’ posts were negative. The journalists have a broader perspective and a better understanding of societal and judicial issues, they express themselves accordingly, and the artificial intelligence entity is capable of parsing such sentiments manifested in social media. Hopefully, the time will come when also more political, judicial and civil participants fully understand the main issues in Romania, as journalistic texts also reflect them on social media.