Analysis of the Problem of Detecting Public Opinion Trends in Ukrainian-Language Posts by Means of Clustering and Neural Networks

In this study, we employ a hybrid clustering approach combining DBSCAN and K-means algorithms on vectorized Ukrainian social media posts to detect public opinion trends. The methodology uses a neural network–based multilingual text vectorization model to capture semantic content effectively. Experiments on a dataset of 90 Ukrainian posts (collected March–May 2025) identified six principal thematic clusters representing key discussion topics. The results demonstrate the effectiveness of the proposed method for social media trend analysis and its potential application in public opinion monitoring.