Development of Sentiment Analysis Algorithms of Uzbek Patient Reviews

The presented work describes the training of three machine learning models (Logistic Regression, Support Vector Machine, Naive Bayes) for sentiment analysis of Uzbek user reviews in the field of medicine. The training data was a corpus of more than <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{5, 0 0 0}$</tex> reviews in the Uzbek language, each of which was assigned to one of three classes: “positive”, “negative” and “neutral”. To objectively assess the effectiveness of each model, experiments were conducted, where, according to the results of the experiments, the naive Bayes classifier showed the lowest accuracy and recall, while logistic regression and the support vector machine were much more stable. In particular, the support vector machine demonstrated a slight but stable superiority in all key metrics (Precision, Recall, F1). At the same time, the authors conducted a comparative analysis of existing solutions, and also provided the necessary information on the Uzbek language, its nature and morphology in general. Language difficulties were noted that could complicate the development of the necessary solution.