Using Machine Learning Algorithms for Kyrgyz Sentiment Analysis

Nowadays, social media platforms, forums and online communities have become central arenas of public discourse where individuals can freely express their opinions, feelings and attitudes. The proliferation of these platforms has led to the generation of massive amounts of unstructured data in multiple languages, providing a unique opportunity for sentiment analysis – a technique for identifying and categorizing opinions expressed in text data. Despite the global reach of sentiment analysis, there remains a significant gap in research focusing on less-researched languages ​​such as Kyrgyz. This study fills this gap by conducting a comprehensive sentiment analysis of Kyrgyz comments on various online platforms. A range of machine learning algorithms have been used, including traditional methods such as K-Nearest Neighbors (KNN) and Naive Bayes (NB), but also more advanced techniques such as Long Short-Term Memory (LSTM) networks and Recurrent Neural Networks (RNNs). Among the evaluated models, logistic regression (LR) was found to be the most effective, achieving the highest accuracy (0.83) and the highest F1 measure (0.84). These results highlight the potential of LR in sentiment analysis tasks for the Kyrgyz language and provide valuable insights in the field of multilingual natural language processing.