Open AccessPROCEDURE OF ASSESSING CREDIBILITY TO THE EXPERT IN TERMS OF THE PROBLEM OF AGE CATEGORIZATION OF WEB CONTENT
Author(s) -
Valery Vladimirovich Davidyuk,
Askhat Zulpukharovich Baitumenov,
Valeriya Andreevna Gostyunina,
Nadezhda Valerievna Daviduk
Publication year - 2019
Publication title -
vestnik astrahanskogo gosudarstvennogo tehničeskogo universiteta. seriâ: upravlenie, vyčislitelʹnaâ tehnika i informatika
Language(s) - English
Resource type - Journals
eISSN - 2224-9761
pISSN - 2072-9502
DOI - 10.24143/2072-9502-2019-2-86-97
Subject(s) - credibility , reputation , computer science , categorization , set (abstract data type) , quality (philosophy) , expert system , duration (music) , information retrieval , data science , artificial intelligence , political science , law , art , social science , philosophy , literature , epistemology , sociology , programming language
Currently the solution of a set of tasks is impossible without using expert procedures. At the same time, some of them require the involvement of the great number of experts, increasing the probability of human factor errors, fabrication and falsification of expert assessments. There has been suggested the method of obtaining the quantitative assessment of the degree of trust to experts, which allows to improve the quality and accuracy of the expert procedures. The developed method helps to consider both the established reputation of the expert and his reaction in the course of the expertise. The description and examples of calculations of key indicators involved in the formation of the expert's feedback coefficient (in applying one of the most common expert methods - a survey), such as: duration of filling/answering a question, time difference between answers to the related questions, availability of incorrect answers to trap questions. The adequacy of the method has been confirmed by the results of the computational experiment conducted in a group of 20 experts solving the problem of web resource age categorization. The developed method is amenable to further algorithmization and can be applied in decision support systems for various purposes.