METHOD FOR RANKING THE RELIABILITY FACTORS OF TEXT MESSAGES

Abstract

In recent years, the problem of information reliability has become a subject of increased scientific interest, which led to the formation of an interdisciplinary approach that combines methodologies of computer science, psychology, sociology and media education. Within the framework of the modern scientific paradigm, a comprehensive study of various aspects of this subject is carried out: from natural language processing and analysis of fake news to the study of the mechanisms of disinformation spread in social networks, the features of functioning of scientific and political communication, as well as manifestations of information confrontation.

Despite the existing scientific achievements, the proposed study presents the initial phase of developing a new concept that involves the use of factor analysis to solve problems related to assessing the reliability of information messages. The main idea is in transition from an a posteriori to an a priori approach, which allows for a predictive assessment of the data reliability even before their potential appearance in the information space. Within the framework of the proposed model, a set of factors is identified that influence the reliability level of text messages. To arrange the specified factors according to their significance degree, a ranking method is used in combination with semantic modelling based on the language of predicates, which provides a linguistic interpretation of the relationships between the elements of the system. Taking into account the expert determination of weight coefficients for the types of relationships in the semantic network, preliminary weight priorities for each factor are established. Based on the mathematical formalism of the algorithm for calculating the weight of predicates, a generalized description of weight sets for relationships of the “influence” and “dependency” types is constructed. This allows forming a system of integrated weight preferences that determine the factor priority levels in terms of their influence on the reliability of information messages. As a result, a multilevel model of factor influence is constructed taking into account the additional action of predicates that reflect semantic relationships in the information space.