The problem of development of an effective method for text authorship identification (on the material of publications of well-known Ukrainian journalists) is explored. Most existing methods require text preprocessing, which entails new costs when solving the set problem. In the case where the number of possible authors can be minimized, this approach is often excessive. Another disadvantage of the existing approaches is that their vast majority was applied to texts in foreign languages and did not take into consideration the peculiarities of the Ukrainian language. Therefore, it was decided to develop an approach that makes it possible to identify the author of the text in Ukrainian without preprocessing and give high accuracy results, as well as to establish what types of artificial neural networks provide the minimum error for Ukrainian publicists.The developed method uses a multilayer perceptron of direct distribution, the algorithm of supervised learning, vectorization HashingVectorizer, and Adam optimizer. It was determined that with a small number of iterations (4–5 iterations) of artificial neural network learning, we obtain a rather high accuracy of identification of authorship of journalistic texts and rather small value of error. Over 1,000 fragments of texts by three Ukrainian authors were used. As a result of the conducted experiments, it was found that the application of the developed approach to solving the set problem enables achieving rather high results. In the texts containing not less than 500 characters, the accuracy reaches 91 %, and the maximum number of iterations of artificial neural network learning does not exceed 15. Such results were achieved primarily due to the efficient selection of the vectorization method at the preparatory stage and the structure of an artificial neural network
With the development of technology, new methods are being created to solve various tasks. Methods that use artificial neural networks (ANN) have become part of our routine and are very rapidly rooted in the Internet and computers. This work uses modern achievements in the ANN and methods of their training to solve the problem of correlations between objects, namely, vacancies and candidate information. The task of recruiting staff to expand or upgrade their staff has been and still is today. With the development of the Internet, this task has made new sense, because many resumes are scattered across different web services. So the question is: How do employers review and process information on all job applicants? This is what artificial neural networks will help, because with their help this process can be completed several times faster.Methods that use artificial neural networks (ANN) have become part of our routine and are very rapidly rooted in the Internet and computers. This work uses modern achievements in the ANN and methods of their training to solve the problem of correlations between the objects, namely a vacancy and information about a candidate. A vast array of data is processed and relations between the specific components are discovered. The suggested approach can be used not only for this specific task but also for a whole class of similar tasks like classification and pattern recognitionAll these tasks can be solved with different ANN parameters which should be configured for every particular set of data. There are many ways to improves. From input text normalization and different vectorization algorithms to ANN topology, parameters, and training algorithms. Considered ANN and its parameters with chosen training algorithm works with high accuracy level only for tasks described in this paper. for every other task, there is a need to experimentally and theoretically proven other parameters and topology to reach high accuracy.
The study proposes the approach to identifying whether a text belongs to certain media with the help of the support vector machine. Based on the results of studying the texts of the Ukrainian media “Ukrainska Pravda” and “Puliteka”, it was determined that, depending on a change in parameters or machine learning architecture, the result improves or worsens. It has been established that on the basis of a database containing 6000 text fragments, evenly distributed in alphabetical order, the reliability coefficient of identification of texts as belonging to the specific mass media using a SVM (SVC / SVR) is about 0.99. It has been determined that at the lexical language level the text is defined as belonging to certain media, mainly due to lexico-semantic and, to a lesser extent, lexico-stylistic peculiarities.
Ensuring the accuracy of biomedical information is critical. Traditional validation methods are slow and error-prone, making automated solutions increasingly necessary. This study introduces LLM-FactCheck, a tool that is used to improve data quality in SemMedDB, a large biomedical database. Using LLMs such as Llama, GPT-4.0, GPT-3.5 Turbo, and Mistral, the system validates biomedical statements of subject-predicate-object triples. This approach increases the reliability of biomedical research by removing erroneous data input, ensuring that only accurate information is retained for further analysis. Comparisons of different LLMs show significant improvements in accuracy and efficiency. This research represents a significant step forward in automated fact-checking to improve the integrity of biomedical databases and support informed clinical and research decisions. The tool and additional resources related to this study can be found in the GitHub repository at https://github.com/KoslickiLab/LLMFactCheck.