Software for a computer system for recognizing authors of texts has been developed
18.02.2025 /
Nowadays, an increasing number of studies on the processing of texts written in natural language contribute to the solution of current problems in this area. One such area is the categorization of texts. Texts can be categorized into different categories based on subject matter, sentimentality and writing style. Categorization of texts according to writing styles is widely used to identify genres of documents and to recognize authors of texts. Identification of text authors is done on the basis of quantitative analysis of writing styles of text authors - writing stylistics.
Various scientific institutions of Azerbaijan, including scientists of the Institute of Control Systems (ICS) of the Ministry of Science and Education of the Republic of Azerbaijan, contribute to this research, which is carried out on a global scale. One of the important scientific results obtained in the Institute for 2024 was the work “Development of software on the basis of mathematical models using various methods of machine learning - artificial neural networks, methods of reference vectors - for computer system of recognition of authorship of texts”, carried out by the researchers of the laboratory “Methods of recognition, identification and optimal decision making” under the guidance of corresponding member of ANAS, Professor Kamil Ayda-zade, researcher Rustam Azimov, PhD in Engineering, Associate Professor Elshan Mustafayev and software engineer Sona Rzayeva. The main purpose of the research is to study the issues of recognizing authorship of texts in Azerbaijani language, development of solution algorithms and computer system.
Text authorship analysis is widely used to identify the author of an anonymous or disputed work of art (especially in copyright litigation), to verify the authenticity of suicide letters, for intelligence purposes (e.g., to determine whether anonymous messages or statements were written by a known person), to identify the author of malicious computer programs (e.g., computer viruses), and to identify the authors of some texts on the Internet (emails, blog posts, texts on the Internet).
The software modules for structural and parametric identification of artificial neural networks used in the study and the results obtained were used to improve the application software package “Form Recognition System” in the State Examination Center of the Republic of Azerbaijan. The results of the work have been published in 2 impact factor journals included in Web of Science database and 3 journals included in Scopus database.
