Accepted_test

In the modern world, characterized by wide access to internet resources, the average person consumes a vast amount of information daily, including both accurate data and false statements. A particular problem is posed by manipulative texts aimed at distorting the perception and mood of the reader. Recognizing such text online is more challenging than identifying manipulation in personal communication, which is why we propose a virtual assistant capable of detecting potential manipulation based on the analysis of cognitive distortions in text communication. Thus, the algorithm presented in this work can be used for monitoring public health based on public communications. Additionally, the last century has seen a significant increase in the quantity and quality of psychological research, which, combined with the accessibility of information on the internet, has led to a heightened interest in psychology and an increased burden on psychotherapists. Our algorithm for finding cognitive distortions in text can help the specialist optimize diagnosis and subsequently monitor the effectiveness of therapy.

The algorithm analyzes texts based on pre-created dictionaries in English, containing key words and phrases for each type of cognitive distortion. Using the n-gram method and prioritization by size, it assesses the expression of distortions and assigns each a coefficient from 0 to 1. We researched the algorithm's parameters on two datasets and determined their optimal values.