UDC 007:681.512.2
NEWS RELIABILITY IDENTIFICATION USING ML-MODELS
I. Yu. Kashirin, Dr. Sc. (Tech.), full professor, RSREU, Ryazan, Russia;
orcid.org/0000-0003-1694-7410, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The article discusses a new technology for designing trainable artificial intelligence models, designed to assess the veracity of electronic news materials. Trained models with knowledge (Machine Learning, MLmodels) are used. Knowledge structures are implemented in the form of semantic networks that describe initially uncalculated input data elements. Вестник РГРТУ. 2023. № 83 / Vestnik of RSREU. 2023. No 83 47 The technology uses production expert rules to determine the index of reliability of the facts presented in the materials of a news article. Each of expert rules requires the creation of an appropriate program module, most often based on the methodology of syntactic and semantic analysis of natural language. The calculated indices are used as S-elements of neural networks or as input features for training, testing and validation of ML models. The knowledge base of program modules contains rating characteristics of electronic publications and ratings of the authors of news articles. The experimental part of the research was carried out for test software implemented in Python v.4 (Anaconda 4). As source texts for news articles the materials from international Kaggle repository and news feed of Russian mail.ru e-mail service were used. The performed series of experiments makes it possible to evaluate the technology under consideration as a technology for assessing the reliability of natural language texts, which is not inferior in efficiency to international analogues available today. The aim of the work is to create an original technology for automated analysis of natural language news texts published in electronic web resources for the reliability of the information contained in them.
Key words: fake news, reliability of information, ML models, data mining, knowledge base, semantic networks, natural language analysis, production rules