PETRYCHKO M. Information technology for reviewer assignment problem in the field of research expertise

Ph.D. thesis: 207 p., 10 tables, 62 figures, 5 appendixes, 172 references. INFORMATION TECHNOLOGY FOR REVIEWER ASSIGNMENT PROBLEM IN THE FIELD OF RESEARCH EXPERTISE The goal is to increase the efficiency of the management of the processes of scientific works peer review due to the development of information technology that allows automating the process of assigning relevant reviewers. The object of the research is decision support information systems for managing the expertise of scientific activity. The subject of the research is models and algorithms for text processing from the profiles of scientific works and reviewers, and decision-making about the assignment of reviewers. Research methods. The research used the methods of the theory of relational databases to formalize models and operations on reviewers' data and scientific works; methods of applied statistics, computer linguistics and intelligent data analysis; methods of object-oriented programming; discrete optimization methods; methods of experiment planning. The key stage of scientific peer review is the assignment of reviewers according to the subject of the work. Taking into account the constant growth of the number of scientific works and the need to shorten the terms of review to save time and resources, it is advisable to automate the selection of reviewers as much as possible. At the same time, it is necessary to ensure: 1) timeliness of the assignments of reviewers; 2) independence from the authors of the reviewed work; 3) compliance of the reviewer with the topic of the work and many other restrictions and requirements. The complexity of the automatic assignment of reviewers is due to the need to process unstructured or weakly structured information, in particular, various natural language texts generated by reviewers and authors of reviewed works. Based on the obtained scientific results, an information technology was developed that automates the complex and time-consuming process of assigning reviewers using a limited set of initial information. In addition, the levels of kinship of research specialties within the Australian-New Zealand ANZSRC-2020 research classification system and the level of kinship of Ukrainian educational specialties have been identified, which has separate practical significance for solving such problems of managing educational and scientific activities as correcting the list of specialties and fields of knowledge, tracking trends in interdisciplinary research, forming promising interdisciplinary educational programs and competitive proposals for a wide competition, modernization of the network of faculties and universities. Datasets for the categorization of scientists and the assignment of reviewers have also been created, which has a separate practical value for testing and improving new algorithms for processing such information. On the formed dataset for the assignment of reviewers, as a result, it was possible to improve the composition of one-time dissertation councils, when compared with councils from the institution, by an average of 32-55%, depending on the type of algorithm used.