Technology of Construction of Expert Information Web System of Identification and Verification of Priority Ecological Problems in Water Bodies of the River Basin
В. Б. Мокін, А. М. Лучко, O.M. Davydiuk, Т. Є. Вуж · Visnyk of Vinnytsia Politechnical Institute · 2021
The article considers the collection, verification and generalization of a large number of expert assessments of the current state of waters, existing environmental problems and influential factors that increase the risk of failure to achieve environmental goals of each volume of water during the development of river basin management plans (RBMPs) or stabilization of good ecological status of water in the water bodies of this basin. The task is complicated by the large number of such volumes of water, as it is extremely difficult to gather reliable information about the objects located in each of them. Creation of a web system with a map of water volumes and the involvement of a large number of experts from among locals who are not indifferent to the problems of their environment will help to solve this problem. However, then there is the problem of verifying the assessments of these experts, taking into account their different qualifications, experience, different views of the objectives of the RBMP, and the problem of how to compare them to identify the most vulnerable regions by different criteria. It is proposed that experts will be required not only expert assessments on the basis of single directories of possible answers, but also - links to text web resources that confirm their assessments to solve this problem. And then it will be analyzed if these sources really confirm each assessment of the appropriate type of problem for a given region. The authors consider various approaches for comparing expert assessments both based on the basis of fuzzy sets and with the help of machine learning and natural language processing (NLP) technologies. Analogues of the system developed by the authors are considered. A method has been developed to identify and verify priority of environmental problems in water bodies of the river basin based on fuzzy expert estimates, taking into account the probabilities that the text materials cited by the expert do correspond to this problem. These probabilities are determined to use models of NLP technologies. The stages of functioning of the expert information web system for the implementation of the proposed technology are described, which will simultaneously collect the most reliable and detailed information about the objects of water bodies and accelerate its processing and ranking. An example of the implementation of an information web system for identifying priority environmental problems in the water bodies of the Southern Bug River basin is given. Examples of calculating the reliability of expert estimates using the author’s program in Python based on NLP-model BERT and logistic regression which were applied to real text information are given.