Identification of Fake News Using Emotional Profiling as an Approach to Text Analysis

This article presents new methods and tools used in the field of text analysis to identify fake news in the media. The problem with the research is that, as a rule, to identify fakes, a training dataset is required, on which thematic fakes were tested. This is not always feasible and requires additional resources. To solve this problem, a comprehensive research methodology has been developed that covers most detection tools, even in the absence of an established database containing reliable and fake news. The study includes a combination of various algorithms combined into a single analytical structure, presented in the work in the form of pseudocode. The authors introduce the concept of an emotional fake model similar to individual emotions included in a broader emotional spectrum. The essence of the model is to evaluate fakes based on the structure of definitions of emotions formed in fakes, which differ from the original signals due to different weight coefficients. The innovation involves a two—stage identification of fakes - initially clusters of messages from the text corpus are identified, and then, based on text analysis tools, their linguistic features and emotional differences are revealed (based on a set of emotions POMS). In the context of creating fake news using neural networks, emotional coloring plays a crucial role, providing a permanent foundation that can serve as a cornerstone for identification.