A Socratic Principle-Based MethodFor Fighting Fake News

Abstract

Misinformation and fake news are a typical problem for media that can lead to negative consequences in society. This work represents research into a solution that indirectly fights misinformation by introducing different perspectives through opinions. The goal is to crumble information bubbles and narrowed-minded views and create a neutral information environment. We developed a system that searches and returns different opinionated documents given a reference document. The system extracts so-called sentimented topics that hold information about the sentiment score for a certain topic by utilizing topic modeling and sentiment analysis. Additionally, we introduce a search algorithm to find the best combination of k sentimented topics that creates a neutral informative setting based on distance values.

To analyze the system, we used a collection of text documents related to the subject COVID-19, which we extracted from the web. We evaluated the performance of the proposed system and the quality of the output of the system. Results showed that the components responsible for preprocessing and selection of sentimented topics are computationally expensive. The proposed system does return text documents that are both related and bear different sentimented topics compared to the reference document. Although the system has some flaws, we believe this system could serve as the basis on which to create a neutral informative setting.