Information system for determining the polarity score of a sentiment in Ukrainian-language content

Abstract

The article discusses the features of implementing a naive rule-based method for determining the polarity of emotional coloring in Ukrainian texts. This approach allows you to quickly assess whether the tone of the content is positive, negative, or neutral. The method includes preprocessing of text data: removal of emojis, links, hashtags, and special characters, as well as tokenization and lemmatization. Next, a polarity dictionary is used, in which each word has a specific rating, allowing the overall polarity of the text to be calculated. The paper provides an example of applying the method to a test dataset and demonstrates the effectiveness of the system in automatically determining content polarity.